C $Id$
0 mosflm Help Routines

 0 mosflm Help Routines

1 @FILENAME

   Commands (ie keyworded input) for MOSFLM will be  read  from
   the file "FILENAME".  This is particularly useful when running online.

1 ADCOFFSET 

  ADCOFFSET <IOFF>

    This defines the adc offset (IOFF) of the detector. Its precise
    value will only be critical when processing images with a very low
    background.
    A reasonable estimate can be obtained by examining an image with no
    exposure to X-rays.
 
   Default 8 or 4 for Raxis IV detectors or 1 for Fuji scanners

1 ADDPART

   ADDPART 

   For partial reflections extending over only two images, this will
   sum the pixel values for the two halves of a partial within MOSFLM,
   and the resulting reflections are treated as fully recorded.

   Summed partials will be used in the formation of the standard 
   profiles.

   The use of ADDPART is only appropriate if every image has had the
   same effective exposure time. It should not be used if the incident
   beam intensity changes significantly during one image (eg at a
   synchrotron).  ADDPART is NOT in general appropriate for use with
   images collected on an offline scanner, or if the position and
   orientation of the image wrt the camera is not identical for every
   image. In particular, if the Mar scanner is not locking properly
   leading to a variation in the ROFF parameter then ADDPART should
   not be used. When ADDPART is used inappropriately it is likely that
   a lot of reflections will be flagged as badspots because of a high
   PKRATIO (ie a poor profile fit).

   The ADDPART option has been made largely obsolete by the new
   scaling program SCALA, which allows the use of partials in scaling.

   Default: Partials are NOT added together within MOSFLM


1 ALLOUT

   ALLOUT

   If this keyword is present, then all reflections except those
   within the cusp or outside the detector limits or outside the
   resolution limits will be written to the output MTZ file. This is
   for special use only.

   Default: Badspots, spots flagged as overlapping (and overloads
   unless explicitly requested) are omitted from the output MTZ file.

1 AUTOINDEX

   AUTOINDEX [ THRESHOLD <ithresh> ] [ IMAGES < id1 id2 ...> ]
             [ PHI <phistart phiend> ] [ IDENT <ident> ]
             [ TEMPLATE <template> ][ FIXCELL ] [ UNFIXDIST ] 
             [ DPS <EXCLUDE <ICE> <AUTO> < rmin rmax ...>> 
                   <EXTRA> <FILE filename> <MAXCELL xxxx> <MINCELL xxxx> 
                   <PREREFINE> <REFINE> <SAVE> <SDCUTOFF> <SOLUTION n> 

   Invokes the autoindexing routine when running a background job.
   (Autoindexing can also be done interactively using the menu in the
   X-windows display...this is invoked by giving an IMAGE keyword). If
   IMAGE keywords have been given, then spots from those images will
   be used in the autoindexing. If no IMAGE keyword is given, then the
   first image from the PROCESS keyword will be used. 

   The DPS keyword allows indexing in a background job when the
   spacegroup and cell are unknown. Thus if the AUTOINDEX keyword is 
   given without the DPS keyword, either a MATRIX or a CELL and UMATRIX
   keywords plus a SYMMETRY keyword MUST be given.
 
   By default, cell parameters are refined during autoindexing but the
   crystal to detector distance is fixed. However, the cell can be
   fixed using the FIXCELL keywords respectively. Under these
   circumstances it may be appropriate to refine the detector
   distance. This can be done using the UNFIXDIST keyword. Note that
   it is NOT a good idea to refine both cell parameters and the
   detector distance as these are highly correlated.

   Note that if several AUTOINDEX keywords are given, FIXCELL and
   UNFIXDIST must be given on the LAST of the AUTOINDEX keywords, or
   they will not take effect.
P
   Examples:

   AUTOINDEX                           ! Autoindex using first image ONLY

   AUTOINDEX IMAGES 1 2 3              ! Autoindex using first three images

   AUTOINDEX IMAGE 1 PHI 0 1 IMAGE 20 PHI 50 51 IDENT test_2
                                       ! Autoindex using image 1, phi range
                                       !  0-1, and image 20, phi range 50-51
                                       ! and image identifier "test_2".

   AUTOINDEX FIXCELL                   ! Autoindex but keep the cell fixed.
                                       !
   AUTOINDEX DPS IMAGE 1               ! Autoindex using the DPS indexing code
                                       ! with the first image ONLY.

   AUTOINDEX DPS IMAGE 2 MAXCELL 120   ! Autoindex using the DPS indexing code
                                       ! with image 2 and a maximum cell edge 
                                       ! length of 120 Angstroms

   AUTOINDEX DPS IMAGE 2 20 REFINE     ! Autoindex using the DPS indexing code
                                       ! with images 2 and 20 and refine all the
                                       ! solutions with a penalty < 200

   AUTOINDEX DPS IMAGE 5 SOLUTION 10   ! Autoindex using the DPS indexing code
                                       ! with image 5 and pick the solution 
                                       ! ranked no. 10 in the list
p

2 THRESHOLD
  THRESHOLD <ithresh>

   Only spots with an intensity greater than <ithresh> times its
   standard deviation are included in the autoindexing.

   Default 20.

2 IMAGES
  IMAGE < id1 id2 id3 ...>

  Image numbers id1,id2 id3 etc will be used in autoindexing. Up to a
  maximum of twenty images can be specified.

  Default: The first image to be processed will be used for autoindexing.

2 PHI
  PHI <phistart phiend>

  The start and end phi values for the image specified on the
  preceeding IMAGE keyword. When the PHI subkeyword is given, only ONE
  image may be given on each IMAGE sub. keyword.

  Default: The phi values for the images are calculated from the phi
  value for the first image and the oscillation angle specified on the
  PROCESS keyword, ie it assumes all images to be used are part of a
  continuous phi rotation with the same oscillation angle for each
  image.


2 IDENT
  IDENT <identifier>

  This sets the identifier for the image filename for the image
  specified on the preceeding IMAGE subkeyword. When the IDENT
  subkeyword is given, only ONE image may be given on each IMAGE

  Default: Use the image identifier specified on the IDENT keyword.

2 TEMPLATE
  TEMPLATE <template>

  This sets the template for the image filename for the image
  specified on the preceeding IMAGE subkeyword. When the TEMPLATE
  subkeyword is given, only ONE image may be given on each IMAGE

  Default: Use the image template specified on the TEMPLATE keyword.

2 FIXCELL
  FIXCELL

  Do not refine the cell following autoindexing, i.e. keep the user-input
  cell parameters (must be given on the last AUTOINDEX keyword in a run).2 

2 DPS
  DPS <MAXCELL REFINE SOLUTION EXCLUDE PREREFINE EXTRA FILE >

  Use the DPS autoindexing routines in background. This is the only way to
  invoke indexing from an unknown cell in a background job.

3 EXCLUDE
  EXCLUDE <ICE> <AUTO> < rmin rmax >

  Excludes resolution ranges of spots from the autoindexing. Specifically;
  'EXCLUDE ICE' will exclude spots near the ice-rings at ~3.90, ~3.67, 
  ~3.44, ~2.67, ~2.25 and ~2.05A. 

  'EXCLUDE AUTO' will attempt to find rings of peaks in the spot list, and
  will then exclude spots near these rings. This is untested and not 
  recommended.

  'EXCLUDE rmin rmax' will exclude any spots within the specified resolution
  ranges.

  Of course, these are critically dependent on the data collection 
  parameters being correct, especially the crystal to detector distance, the
  direct beam coordinates, and the wavelength.

  Default: no spots are excluded.

3 EXTRA
  EXTRA <ON>

  Extra information in the form of a distortion index is printed along with the 
  penalty. Anything other than EXTRA ON turns this off.

  Default: no extra information is printed. 

3 FILE
  FILE <filename>

  Read the spot list from a file. This may be useful for use with third-party
  spot finding routines, but is also employed in the new GUI.

3 MAXCELL
  MAXCELL <n>

  Impose a maximum cell edge of n Angstroms for the DPS autoindexing. 

  Default: determined by the program from a consideration of the spot 
  separation on the image(s).

3 MINCELL
  MINCELL <n>

  Impose a minimum cell edge of n Angstroms for the search of the FFT
  in DPS autoindexing. Note that cell edges shorter than this may still
  be returned as a result of cell reduction.

  Default: 0 (zero) Angstroms.

3 PREREFINE
  PREREFINE <ON/OFF>

  Refine the triclinic solution before applying lattice symmetry. This usually
  improves autoindexing, but if there are few spots or the basis solution is
  incorrect the refinement can be unstable.

  Default: prerefinement is performed.

3 REFINE
  REFINE

  Refine each solution (after applying lattice symmetry) with a penalty less 
  than 200 before choosing.

  Default: Not to refine.

3 SAVE
  SAVE

  Save the results of indexing in a file for later use.

3 SDCUTOFF
  SDCUTOFF <n>

  Exclude spots which shift by more than <n> SDs during the cell refinement.

  Default: 2.5 sigma

3 SOLUTION 
  SOLUTION <n>

  Tell the program to choose solution number <n> in the list. 

  If <n> = 0, Mosflm will stop and wait for more user input before applying 
  symmetry and refining the cell. A subsequent AUTOINDEX DPS SOLUTION <n> 
  (where <n> is from 1 to 44) is required to pick the desired solution; this 
  can be done multiple times. 

  Default: To choose the highest symmetry Bravais lattice with the biggest
  penalty below an obvious cutoff. 

2 UNFIXDIST
  UNFIXDIST

  Allow the crystal to detector distance to refine following autoindexing.
  It is recommended that this is only used with the FIXCELL keyword
  (must be given on the last AUTOINDEX keyword in a run).


1 AUTOMATCH
  Use the automatic pattern matching procedure to refine the 
  missetting angles. Use the refined values to update 
  parameters on generate file and then create a new reflection
  list and measure the image (unless NOMEAS sub-keyword given).

   Note that ONLY a single pack per generate file can be processed
   with AUTO, unless POSTREF is also specified. In this case,
   AUTO refinement will only be done on the first pack, but
   post refinement will be done on all packs.

  Type More for more info.

2 More
  An automatic pattern matching algorithm which is basically
  that described by Rossmann (J. Appl. Cryst. (1979) 12,
  225-238) has been implemented. First, the positional
  refinement of the central region of the image (CENTRS),
  is carried out. If a large radius of convergence has been
  requested, the reflection list used to locate reflections
  to be used in the refinement will be generated with an
  increased mosaic spread. (This will be determined by the
  values of RCONV, OVERLAP and the oscillation range of the
  photograph). Both fully recorded and partially recorded
  reflections will be used in the refinement, since at this
  stage the two cannot be distinguished. This can result
  in higher than usual positional residuals. In this refinement,
  the parameter ccomega is 100% correlated with the missetting
  angle PSIX (which corresponds to rotation around the X-ray beam).
  Therefore, following refinement, PSIX updated by the value
  of (ccomega-true_ccomega) and ccomega is reset to true_ccomega
  where true_ccomega can be supplied by the user (sub keyword
  CCOMEGA) and defaults to zero.

  Following successfull CENTRS refinement, an initial "biglist" 
  of reflections is generated using an artifically high value 
  for the mosaic spread. The value used is determined by the 
  radius of convergence requested by the user (RCONV). The 
  integrated intensities of all reflections in this "biglist"
  are measured, including overloads. 
    The PSIZ missetting angle is then varied in a number of
  steps (NSTEP) from -RCONV to +RCONV about its input value.
  For each step a new reflection list is calculated (using the
  correct mosaic spread) and the total intensity evaluated by
  comparison with the "biglist". The step which gives the
  greatest intensity is chosen as the starting point for the
  a second test in which only the intensity LOST is monitored.
  The two steps (one in positive and one in negative directions)
  at which the intensity has dropped by more than ten standard
  deviations are found. The optimum angle is chosen as the
  midpoint of these two steps. This procedure has been found to
  be much more effective than simply searching for the step
  giving the greatest intensity, particularly for very strongly
  exposed images where there is significant diffuse scattering at
  Bragg positions for reflections not actually recorded on the
  image.
    The PSIY angle is then refined in the same way. This 
  constitutes a single pass. More than one pass can be made
  (NPASS), and on each pass the stepsize is multiplied by
  the parameter DAMP. If, after anf pass, the change in PSIX
  or PSIY is greater than 0.5*RCONV or 0.75 degrees (whichever
  is the smaller), or the sum of the shifts in all three angles
  is greater than 0.5*RCONV or 1.5 degrees (whichever is the 
  smaller), then the positional refinement (CNETRS) is repeated
  and the "biglist" regenerated and remeasured using the new
  orientation.                                 
    Following completion of one round of refinement, the entire
  procedure may be repeated at higher resolution if requested
  on keyword RESOL.
    On completion of orientation refinement, the new missetting
  angles are written back to the generate file, a new reflection
  list calculated (out to the full resolution limit)
  and the image is measured in the normal way.

2 RESOL
  RESOL <x1,[x2]>
  Reflections out to resolution limit x1 Angstroms will be
  used in pattern matching. If x2 is given, the refinement
  will be repeated at the resolution x2. A lower resolution
  limit gives a larger radius of convergence, but a higher
  limit will give more accurate missetting angles. The best
  values will depend on the cell parameters, but for about
  100A cell, 6A and then 4A works well.
   Default  x 6A

2 NPASS
  NPASS <I1>
  Specifies the number of cycles of PSIY,PSIZ refinement at
  each resolution.
  Default 2

2 RCONV
  RCONV <X>
  The required radius of convergence (degrees). This value
  (together with NSTEP and DAMP) determines the stepsize
  used in the refinement of PSIY,PSIZ. The stepsize is
  RCONV/NSTEP on the first pass, and multiplied by DAMP on
  each successive pass. At the second resolution x2 (if
  specified) RCONV is set to twice the stepsize used in the
  LAST pass of the refinement at the lower resolution.
  Default 1.0

2 DAMP
  DAMP <X>
  The damping factor applied to the stepsize used in PSIY,PSIZ
  refinement on successive cycles.
  Default 0.25

2 NSTEP
  NSTEP <I1>
  The number of steps to be used in the stepwise refinement
  of PSIY,PSIZ. This number, together with RCONV, determines
  how large the initial stepsize is.
  Default 4

2 NOMEAS
  Do not measure the image after orientation refinement.

2 ANGLE
  ANGLE <X>
  When refining PSIY,PSIZ, only those reflections lying in a
  sector of half-angle "X" centred on a line normal to the
  rotation axis will be included in the pattern matching. 
  (In practice this doesn't seem to make any difference !)
  Default 90.0

2 OVERLAP
  OVERLAP <X>
  Sets minimum acceptable overlap (in degrees) between the
  oscillation photograph and the initial orientation assuming
  the maximum possible error of RCONV. If there is not enough
  overlap, an enlarged mosaic spread is used in generating
  reflections to be used in the refinement of the central
  region of the image (CENTRS).
  Default 1.0

2 NCYC
  NCYC <I1>
  Sets the number of cycles of central (CENTRS) refinement.
  Default 2

2 RESIDUAL
  RESIDUAL <X>
  Sets the maximum allowed residual for positional refinemnt
  of central region of image (10 micron units). If the
  initial orientation errors are large, then this residual
  can be much larger than is normal (typically 1.0 to 2.0
  for a correctly oriented image)
  Default 25.0

2 ELIMIT
  ELIMIT <X>
  CENTRS refinement is automatically repeated if the initial
  residual is greater then X (10 micron units)
  Default 25.0

2 CCOMEGA 
   CCOMEGA <X>
   Sets the "correct" value of ccomega. This will normally be close
   to zero but for Mar Research image plate scanners it can be as
   large as 0.5 degrees. The missetting angle PSIX is updated by
   the difference between the refined value of ccomega and the 
   "correct" value after the central refinement (See More)
   Default 0.0

1 BACKGROUND

   The background is determined by fitting a least squares plane to
   the background pixels as defined by the measurement box.  To allow
   for outliers in this plane, due to "streaks", dust, cosmic rays or
   other sources of noise, the initial estimate of the background
   plane is calculated using only a fraction (BGFRAC) of the actual
   number of background pixels, and those pixels with the LOWEST pixel
   values are chosen. The background plane constant "c" is adjusted to
   allow for the systematic bias that this would introduce assuming a
   Gaussian distribution of values. Using this background plane, all
   background pixels are tested and those that deviate by more than
   BGSIG times the estimated deviation (which is calculated from
   counting statistics and the value of "c") from this plane are
   rejected as outliers.  The remaining pixels are then used to
   calculate the final plane.  The number of pixels that can be
   rejected is limited at present to a maximum of 400, if there are
   more than this the reflection is flagged as a badspot. The user can
   also set a minimum acceptable number of background pixels for a
   spot (see REJECTION keyword) and a separate number (given as a
   fraction of the total) for refinement spots (see REFINEMENT).

   It should be noted that background pixels which are actually
   part of neighbouring spots are automatically rejected using the
   known positions of these spots and the spot sizes derived from
   the standard profiles.

2 BGSIG
   BGSIG <x>
   Background pixels with values deviating by more than "x" 
   standard deviations from the initial plane are rejected.
   Note that an independant value is used for the standard profiles
   (see PROFILE)
   Default 3.0

2 BGFRAC
   BGFRAC <x>
   A fraction "x" of the total number of background pixels will
   be used in determining the initial plane (but  background pixels
   overlapped by neighbouring spots will also be rejected)

   Default: 0.8

2 RECOVER

  RECOVER <RECLEVEL>

  If there are fewer than (RECLEVEL*NBGMIN) background pixels remaining
  after rejecting those overlapped by neighbouring spots, then those
  overlapped background pixels which have the lowest values in the
  standard profile will be included in background determination, so that
  at least RECLEVEL*NBBGMIN pixels are used.
  RECLEVEL defaults to 5 but can be set by this keyword.
  NBGMIN defaults to 10 but can be reset by keywords: REJECTION MINBG (qv)
  Default 5.0

1 BACKSTOP

  BACKSTOP CENTRE <x y> RADIUS <r>

  This defines the region of the image considered to be behind the
  backstop shadow. Any reflections lying within a circle of radius "r"
  mm centred at coordinates (x,y) mm (with respect to the same origin
  as the direct beam, ie the first pixel in the image) will NOT be
  integrated. This circle is displayed as a red circle in the
  X-window display.

  (This is equivalent to the LIMITS RCENTRE RMIN combination of
  keywords).

  NOTE WELL: If a solid backstop support has been used which casts a
  shadow on the image, an alternative way of dealing with the shadow
  is to use the NULLPIX keyword. See under NULLPIX.

  Default: x,y default to the direct beam position
        r defaults to 0.05*(smaller of maximum X or Y coordinate (MAXX,MAXY))

  
1 BEAM
  BEAM <nimage> [SWUNG_OUT] XBEAM YBEAM 

  XBEAM, YBEAM are the direct beam coordinates (derived from wax rings
  or previous knowledge).

  These are the coordinates (in mm) of the direct beam position with
  respect to an origin at the first pixel in the image. The X
  coordinate is in the slowly changing direction in the image, Y is
  the fast changing direction. For Mar images this is the lower right
  of the image as viewed from behind the detector looking towards the
  source. For RAXIS images, this is the lower left corner of the image
  viewed in the same way.

    If the detector does not scan the centre of the image, the
  simplest way to estimate its position is to record a diffraction
  pattern from a sample that will give powder rings eg paraffin wax.
  The centre can then be measured from these rings using MOSFLM,
  IPDISP or by other software (but beware that there may be
  differences in the coordinate conventions when using other
  software).

  Note that MOSFLM displays the image viewed from the detector looking
  towards the source (cameraman's view), and also that the "fast
  changing" direction in the image is ALWAYS vertical in the display,
  regardless of whether it is vertical or horizontal in the actual
  detector. Thus some images will be rotated by 90 degrees.
        
  Direct beam coordinates can be assigned for individual images with
  multiple BEAM keywords, specifying the image number <nimage>:
P
  eg  BEAM 1 100.3  99.7   ! Beam coords for image number 1
   BEAM 2  99.3 100.8   ! Beam coords for image number 2
   BEAM 3 101.0  99.2   ! Beam coords for image number 3
p
  Note that if specifying multiple BEAM keywords, then the PROCESS
  keyword MUST be given before the BEAM keywords, and direct beam
  coordinates MUST be given for ALL images to be processed. If only a
  few images have a significantly different direct beam position, then
  it is possible to set the position for all images with an initial
  BEAM keyword, then reset those that are different:
P
  BEAM 90.3 90.8
  BEAM 17 91.0 90.0
  BEAM 23 89.0 91.0
  BEAM 27 90.5 89.5
p
  will set the beam coordinates to 90.3,90.8 for ALL images and then
  reset the values for images 17,23 and 27.

  Note that multiple BEAM keywords CANNOT be used with the IMAGE
  keyword, only with the PROCESS keyword.

  When processing data from a detector with a two-theta arm, the
  direct beam coordinates are assumed to be those at two-theta = 0.
  However if the keyword SWUNG_OUT is included, the direct beam
  coordinates are then assumed to be those with two-theta set to the
  value specified on the TWOTHETA keyword.

  Default: Physical centre of the detector for all images.


1 BIAS
    BIAS <IADD>
    
    The value IADD will be added to all pixel values in the image.
    This is to overcome the problem of processing images where the adc
    offset and X-ray background is so low that there are pixels with
    values less than or equal to NULLPIX (see NULLPIX keyword, but
    normally zero) within the scanned region of the image. MOSFLM
    interprets such values to mean that the pixel lies OUTSIDE the
    active region of the detector, and so any spots containing these
    pixels will be rejected. To circumvent this problem use keyword:
    BIAS IADD which will add "IADD" to all pixels. In this case you
    MUST ensure that you do not try to measure spots for which part of
    the measurement box is genuinely outside the scanned area, so it
    is advisable to set the LIMIT RMAX to the true radial limit minus
    the measurement box size: LIMIT RMAX 87 For a small Mar scanner
    with a box size of 3mm (20 pixels). The adcoffset will
    automatically be reset to IADD so that standard deviation
    calculations are correct.

    This keyword should NOT be used for tiled CCD detectors where
    there are small gaps between the tiles, otherwise the program will
    try to integrate spots that fall across these gaps, resulting in
    incorrect intensities.

1 BSWAP
   BSWAP

    Forces byte swapping when reading the image. Normally only needed
    for those images which have no header records from which
    the byte-order can be worked out (this is done automatically for
    Mar, R-axis, and Mac Science images), or if the header records are
    just being skipped (unkown detector types...see DETECTOR keyword).

    Default: For images with header records, the byte order is
    determined by reading the header record. Otherwise it is assumed
    they are written on Vax machines.


1 CAMCON

   USE OF THIS KEYWORD IS TO BE AVOIDED !!!
   
   Give the CORRECT direct beam coordinates on the BEAM keyword,
   rather than correcting them with CCX and CCY.

   CAMCON CCX <x> CCY <x> CCOM <x> 

   Defines camera constants (in mm or degrees) for all images.
   Default values are 0.0

2 CCX,CCY
    Distances (in mm) from the input direct beam coordinates
    and the true direct beam coordinates.

2 CCOM

    CCOMEGA represents a rotation around the X-ray beam direction. It
    is normally assumed that either the detector X or Y axis is
    parallel to the rotation axis. CCOMEGA is the angle by which X (or
    Y) deviates from being exactly parallel. As such, it is a constant
    for any particular detector, and is normally very close to
    zero. 

    However, when refining the crystal orientation, any slippage of
    the crystal around the X-ray beam direction is accounted for by
    refining CCOMEGA. In these circumstances, CCOMEGA represents part
    of the definition of the crystal orientation, and can vary from
    one image to the next.

2 More
    The  "fiducial  constants"  CCX,  CCY,  CCOM  are  refined   during
  measurement  of each image and should be adjusted to the nominal values
  obtained on a series of images. 

1 CELL
  CELL [KEEP] < x x x [x x x] >
     Cell parameters. If only the cell lengths are given, the cell
   angles are worked out from the symmetry.
   The cell must be given when specifying U matrix (UMAT keyword).
   
   If an A matrix has also been supplied, the cell parameters given on
   the CELL card will overide those determined from the A matrix.
   
   If autoindexing is being done, and the KEEP subkeyword is present,
   the cell parameters supplied will override those determined by the
   autoindexing.

   Default: none


1 DEBUG OUTPUT
    
    Various kind of debug output can be obtained, and these are
   controlled by keywords DEBUG, DUMP, BAD, SPOT. There are also some
   options to change the flow of processing (PROCESS,NOREFINE) for
   special cases.

2 NOREFINE
    Turn off all positional refinement during processing. This has
   been used to process a flood field image to test for bias in
   integration procedures.


2 PROCESS
  The program will go straight into processing a  pre-existing
  'SPOTOD.DAT'  file,  without doing any positional refinement
  etc.  A previous job must have  produced  the  'SPOTOD.DAT' 
  file.    Note that the measurement  box
  parameters CANNOT be changed when using this option, because
  they have been used when writing the  original  'SPOTOD.DAT'
  file.
  DEFAULT:  Process image normally.

2 DEBUG
  DEBUG [Subroutine names [number of reflections to be dumped]]

    Turns on debug output in the subroutines specified.  If none
   specified, all debug output will be printed. To get dumps of
   actual pixel values from the subroutine, use  SPOT. For some
   subroutines a number can be given to specify the number of
   reflections to be dumped. eg DEBUG WRMTZ 2000 will print debug
   output for the first 2000 reflections. The default number is
   100. Not all subroutines have this option...a warning message will
   be given if the option is not available.
 
  DEFAULT: No debug output


2 SPOT

  Turns on display of the optical densities in the measurement
  box in those subroutines where the DEBUG flag has been set.

   DEFAULT:  No display of optical densities.

2 BAD

    If this keyword is present with NO subkeywords, then a list of
   reflections will be read from the file assigned to logical name
   SPOTDUMP (default SPOTDUMP.DAT). For each reflection, information
   about box size, background plane constants, integrated and profile
   fitted intensities and sd's will be given, together with dumps of
   the reflection pixel values (after background subtraction), the
   scaled standard profile and the difference between them.

    The indices of up to 50 reflections are read in free format from 
    the file.

    If subkeywords are present, they control which reflections will be
   dumped, according either to their position on the detector
   (IXMIN/IXMAX,IYMIN/IYMAX) or in the list of reflections sorted on
   detector X coordinate (ALL,START). In either case the total number
   of reflections dumped can be controlled (TOTAL).

    The reflections to be dumped can also be specified using the HKL
   subkeyword.
P
   examples:
   BAD IXMIN  100 IXMAX 140 IYMIN 500 IYMAX 800 TOTAL 50 
p
   will dump reflections with pixel coordinates between 100 and 140 in
   X and 500 and 800 in Y up to a total of 50 reflections.
P
   BAD HKL 3 -4 2 HKL 15 -8 41
p
   will dump reflections with indices 3,-4,2 and 15,-8,41
P
   BAD RESO 3.8 3.6 IMIN 18000 
p
   will dump all reflections in the resolution range 3.6-3.8A with a
   profile fitted intensity greater than 18000


3 IXMIN <NXMIN>

    Reflections with detector (pixel) coordinates greater than
    NXMIN will be dumped.

    Default: 0

3 IXMAX <NXMAX>

    Reflections with detector (pixel) coordinates greater than
    NXMAX will be dumped.

    Default: number of pixels in X

3 IYMIN <NYMIN>
    Reflections with detector (pixel) coordinates greater than
    NYMIN will be dumped.

    Default: 0

3 IYMAX <NYMAX>
    Reflections with detector (pixel) coordinates greater than
    NYMAX will be dumped.

    Default: number of pixels in Y.


3 ALL
    All reflections will be dumped (unless IXMIN etc or TOTAL given)

3 START
    START  <NSTART>
    Dumping will start at the "NSTART"'th reflection in the list
    sorted on detector X coordinate.

    Default: 1

3 TOTAL 
    TOTAL <NTOTAL>
    A maximum of NTOTAL reflections will be dumped

    Default: 7500

3 HKL
    HKL <h1 k1 l1> HKL <h2 k2 l2>
    Reflections with indices h1,k1,l1 and h2,k2,l2 will be dumped.

3 IMIN
    IMIN <IDMIN>
    Reflections selected by other keywords will only be dumped if
    their profile fitted intensity exceeds IDMIN.
    Default -100000

3 IMAX
    IMAX <IDMAX>
    Reflections selected by other keywords will only be dumped if
    their profile fitted intensity is less than IDMAX.
    Default 10000000

2 DUMP
   DUMP
   There are a number of options for obtaining additional
   information about particular reflections if they do not seem
   to be processing correctly. These are:

   a) For "badspots", ie those failing the criteria set by keyword
   REJECTION (bgratio,pkratio,gradient, too negative),if the option
   PLOT is given on the REJECTION card the pixel values in the
   measurement box will be printed, but his will ONLY work if
   the detector error factor has been assigned a value (keyword
   DETECTOR).

   b) Specific reflections can be dumped in the same way as badspots
   by using keyword BAD and listing the indices of the reflections
   in free format, one per line, in a file with logical filename
       SPOTDUMP (defaults to SPOTDUMP.DAT)

   c) By using the DUMP keyword, which has a number of subkeywords.
   In this case, selected reflections will have various bits of 
   information printed, as well as dumps of the reflection pixel
   values (after background subtraction), the scaled standard profile
      and the difference between them. The standard profiles themselves
   can also be dumped.

3 SPOT
   SPOT
   The pixel values of each spot will be printed from those
   subroutines which have had the debug output turned on (DEBUG
   keyword).


3 SPOTLIST
   SPOTLIST
   All generated reflections are written to a file with the logical
   name SPOTLIST in the same format as the spotlist written by IMSTILLS.

3 PROFL
   PROFL
   Dump the standard profiles (as number fields rather than single
   character representation)

3 REFL
   REFL <NDUMP>
   Every NDUMP'th reflection will be dumped.
   Default 100

3 IMIN
   IMIN <IDUMP>
   Only reflections with intensity greater than IDUMP will be dumped,
   in addition to the restriction of dumping every NDUMP'th reflection.
   Default 0

3 TOTAL
   TOTAL <MXDUMP>
   A maximum of MXDUMP reflections, satisfying IMIN and NDUMP, will
   be printed (limits output which can be voluminous)
   Default 100
                                                                 
3 BGR
   BGR <BGRLIM>
   Reflections with a background gradient greater than BGRLIM will 
   be printed. This is not affected by NDUMP or IMIN, but is still
   subject to the restriction of a total of MXDUMP refelctions
   being dumped.
   DEFAULT 0.0

3 OVER
   OVER
   Profile fitted overloaded reflections (obtained by putting
   OVERLOADS on PROFILE keyword) will be dumped. This is restricted
   to a total of MXDUMP reflections.

3 EDGE
   EDGE
   Profile fitted reflections which straddle the edge of the detector
   will be dumped, if they have been measured (option EDGE on PROFILE
   keyword). Restricted to total of MXDUMP reflections.
                                                           
3 FULL
   FULL
   Only fully recorded reflections will be dumped.

3 PKREJECT
   PKREJECT <NREJP>
   Dump reflections which have more then NREJP peak pixels rejected
   from the profile fit (rejection criterion set by PROFILE
   WDLIM1,WDLIM2,WDLIM3 keywords, see elsewhere)

1 DETECTOR

   DETECTOR <DETECTOR_TYPE [HORIZONTAL or VERTICAL] [SCANTYPE]
         [REVERSEPHI] [OMEGA]

   See also the SITE keyword.

   This defines the type of detector. 

2 Novel detectors

  It is possible to define the orientation of a novel detector type
  providing the following information is known:

  a) What is the most rapidly changing direction in the stored image
  (horizontal or vertical).

  b) The orientation of the rotation axis (horizontal or vertical).

  c) The direction of rotation (see below).

  d) The position of the first pixel in the stored image.


    First, define the camerman's viewpoint as sitting behind the
  detector, looking towards the X-ray source. FROM THE CAMERAMAN'S
  VIEWPOINT (ie from behind the detector looking towards the X-ray source):

  i) Is the rotation axis horizontal or vertical ?

  ii) Defining the SENSE of the rotation axis (Z) as being from right
  to left if it is horizontal, or from bottom to top if it is vertical,
  and then looking down the rotation axis, ie from +Z towards the
  origin, is the sense of +ve phi rotation of the camera clockwise or
  anticlockwise ?

  These two pieces of information are specified on the ROTATION
  subkeyword:
P
  eg  DETECTOR ROTATION HORIZONTAL CLOCKWISE
  or  DETECTOR ROTATION VERTICAL ANTICLOCKWISE
p
  iii) Are the pixel values in the image file stored as horizontal
  lines of pixels or vertical lines, ie is the most rapidly changing
  pixel coordinate the horizontal coordinate or the vertical coordinate ?

  This is specified on the FAST subkeyword:
P
  eg DETECTOR FAST HORIZONTAL  
  or DETECTOR FAST VERTICAL
p
  iv) Is the first pixel in the file containing the scanned image in
  the lower left (LL), lower right (LR), upper left (UL) or upper
  right (UR)of the image (NB Camerman's view).

  This is specified by the ORIGIN subkeyword
P
  eg DETECTOR ORIGIN LL  
  or DETECTOR ORIGIN UR
p
  For completeness, the definition of the orthogonal axial frame is
  illustrated below. By definition Xd is the "slow" direction in the
  image and Yd is the "fast" direction. For different detectors, Xd,
  Yd will have different orientations in the figure below.
P
                                                Detector
   
                                                          /! UR
                       Y-axis                            / !
                               ^                        /  !
                               !                       /   !
                               !                      /    !
                               !                     /  Xd !
                               !  /              UL /   ^  !
                               ! /                  !   !  !
                               !/       X-ray beam  !   !  !
        Source----------Crystal/--------------------!---/--!---->X-axis
                              /                     !  /   !
                           <-/-                     ! /    !
                            /  \+ve phi             ! Yd  /  LR
                           /   /                    !    /
                          /                         !   /
                       Z-axis                       !  /
                      Rotation                      ! /
                        axis                        !/
                                                  LL                          
                                            
p

  v) Is the detector rectangular or circular ? This is specified with
  the DETECTOR keyword.
P
  eg DETECTOR RECTANGULAR
  or DETECTOR CIRCULAR
p
  It is CRUCIAL than all four subkeywords are supplied, that is
  ROTATION, FAST, ORIGIN and RECTANGULAR/CIRCULAR. The program will
  object if they are not all supplied.

  a complete example might then be:
P
  DETECTOR ROTATION VERTICAL ANTICLOCKWISE FAST HORIZONTAL ORIGIN LR RECTANGULAR
p
  The program still needs to know how the stored values in the image
  should be converted to true "counts". Values are normally stored as
  INTEGER*2, but different detector types use different ways of
  converting this. For Mar scanners, they are treated as unsigned
  integers, ie negative values are converted to the true count by
  adding 65535. This is explicitly coded in subroutines INTPXL,
  INTPXL2 and PUTPXL for Mar, Raxis, Mac Science, Molecular Dynamics
  and Fuji scanners and Princeton (at CHESS), ESRF and ADSC CCD
  detectors. The type of conversion to be used is given by the TYPE
  keyword, and can be MAR, RAXIS, DIP, MD, FUJI, CCD1 (Princeton at
  CHESS), CCD2 (ESRF CCD) or ADSC (ADSC quadrant detector). If a TYPE
  keyword is not given, the values are assumed to be unsigned integers
  in the range 0-65525. Thus a complete detector description might be:
P
  DETECTOR ROTA VERTICAL ANTI FAST HORIZ ORIGIN LR RECT TYPE MAR
p
  It will also be necessary to define the physical limits of the
  detector with the LIMITS keyword (qv) and the size of the image and
  whether there are header records with the SIZE keyword. Additional
  code will be required to make use of information in the header
  records. The pixel size also needs to be given with the PIXEL
  keyword.

  As an example, a large (30cm) Mar scanner could be specified by the
  following keywords:
P
  DETECTOR ROTATION HORIZ ANTI FAST VERT ORIGIN LR CIRCULAR TYPE MAR
  SIZE 2000 2000 HEADER 1
  PIXEL 0.15
  LIMITS XMIN 0 XMAX 150 YMIN 0 YMAX 150 RMAX 150 RSCAN 150
  OVERLOAD CUTOFF 100000
  PROFILE CUTOFF 60000
p
  It may also be necessary to byte-swap the image, depending on the
  type of computer that writes the images and the computer on which
  MOSFLM is being run. For unknown detector types, by default images
  will be byte-swapped if the computer on which the processing is
  being done is "big-ended". (For Mar, Raxis, RaxisIV and MacScience
  detectors the program automatically determines if byte-swapping is
  required by looking at the image header). It is usually very easy to
  tell if byte-swapping is required, for example by looking at pixel
  values behind the backstop shadow. If byteswapping is required, then
  use the keyword BSWAP.

2 Commercial detectors on beamlines

  If a commercial detector is being used on a synchrotron beamline, it
  is not uncommon for the orientation of the instrument to have
  changed (eg using a horizontal rotation axis rather than a vertical
  rotation axis). To correct for this, use the OMEGA keyword. This
  specifies the angle between the detector X-axis (defined to be in
  the slowly varying direction) and the virtual detector X axis
  denoted Xd in the figure shown under "Novel detectors".

  The default values for OMEGA are:
P
  Mar:              90
  Raxis II (vertical rotation axis): 90
  Raxis IV: Vertical rotation axis:  90 
  MacScience Dip Scanners: Horizontal rotation axis: 180
                           Vertical rotation axis:    90
  ESRF CCD:        270
p
  To find the value of OMEGA for any othere detector, include the
  keyword DEBUG CONTROL and the OMEGA value will be printed to the
  logfile.


2 SCANTYPE

   SCANTYPE <FINE or COARSE>

   For RAXIS and MD scanners, SCANTYPE can be FINE (105 micron pixel
   for R-axis, 85.85micron for MD) or COARSE (210 micron pixel for
   R-axis, 172 micron for MD).

   Note that when the R-axis was in use at station 9.6 at Daresbury it
   had a horizontal rather than vertical rotation axis and must be
   specified suing the SITE keyword.qv

   Default: DETECTOR defaults to MAR
            If DETECTOR is RAXIS or MD, SCANTYPE defaults to FINE

   Possible combinations:
P
   DETECTOR RAXIS FINE
   DETECTOR MD COARSE
p
2 OFFLINE

   OFFLINE indicates that the images do not have a fixed origin, and
   therefore the normal type of post-refinement or addition of
   partials is not appropriate (because this assumes that any given
   partially recorded spot will lie on exactly the same pixels on
   successive images, as they normally will for Mar or Raxis scanners).
   However for an offline scanner (eg Molecular Dynamics or Fuji) this
   will no longer be the case, because of differences in the way the
   image plate fits in the cassette or is fed into the scanner. If
   OFFLINE is specified, partials at the end of an oscillation are
   stored in memory and when the next image is being processed they
   are retrieved for use in post-refinement or addition of partials or
   both (depending on what has been requested). This feature can even
   be useful for Mar scanners if there is an instability (eg in ROFF
   or TOFF) in the scanner which would otherwise invalidate addition
   of partials.

   OFFLINE is chosen by default if the detector type is MD or FUJI,
   otherwise the normal type of post-refinement and partial addition
   is used.


2 REVERSEPHI

  If the detector is a standard type but the rotation axis is in the 
  opposite direction to usual, this can be dealt with by using the
  REVERSEPHI keyword.

2 OMEGA

  If the detector is a standard type, but the rotation axis is not in
  the usual orientation, this can  be dealt with by giving the
  OMEGA keyword.

2 DETECTOR_TYPE

   DETECTOR_TYPE can be:
P
   MAR (30cm or 34.5cm (Mar345) diameter Mar detector)
   SMALLMAR (18cm diameter Mar detector).... This is ONLY need when
            running the STRATEGY option without reading any
            images. Otherwise the program automatically determines the
            size of the scanner from the image header.
   MARCCD (Mar Research CCD detector)
   RAXIS (for Raxis II scanners)
   RAXISIV or RAXIS4 (Raxis IV scanners)
   DIP2000  (Mac Science scanner, 20cm plate, 80 micron pixel)
   DIP2020  (Mac Science scanner, 20cm plate, 80 micron pixel)
   DIP2030  (Mac Science scanner, 30cm plate, 100 micron pixel)
   DIP2040 (Mac Science scanner, 40cm plate, 100 micron pixel)
   ESRF [CCD] (for corrected images from ESRF CCD detector)
   MD (Molecular Dynamics offline scanners)
   FUJI [BAS2000 or BAS100] 
   ADSC  for the Quantum4 CCD detector (or Brandeis 2x2).
p
   For Mar scanners, the radius of the scanned area is determined by
   reading the header information. No additional information is
   required. No additional keywords are required to distinguish Mar300
   from Mar345 images, or to distinguish packed from unpacked images.

   For RAXIS and RAXISIV scanners, the default is a vertical rotation
   axis. If the detector has a horizontal rotation axis, then the
   keyword HORIZONTAL must be given immediately after the DETECTOR
   keyword.
P
   eg DETECTOR RAXISIV HORIZONTAL
   or DETECTOR RAXIS HORIZONTAL
p
   For Mac Science DIP scanners, the rotation axis is assumed to be
   horizontal. If it is vertical, then the VERTICAL keyword must be
   given immediately after the detector type.
   eg DETECTOR DIP2030 VERTICAL

   For the Brandeis B4 detector, use "ADSC".


1 DIRECTORY
   DIRECTORY <FDISK>
   The directory containing the images. Up to ten different
   directories can be given, on as many DIRECTORY cards as required.
P
   eg VMS:    scratch_disk0:[andrew.images]
      Unix:   /scr0/andrew/
p
   If NOT running under VMS, if the final "/" has not been
   specified as part of FDISK then it will be added.

   Default: none (ie files are in the directory that the program is
            being run from)

1 DISPERSION
   DISPERSION  <x>
   Wavelength dispersion (delta-lamda)/lamda.
   Default: 2.5E-3 (correct for CuKalpha)
 
1 DISTANCE
  DISTANCE <x>
        Crystal to detector distance (mm).

        Default: Zero, but for Mar Research, Raxis and Mac Science
    scanners the distance will be read from the image headers (or tailers).
 
1 DISTORTION
          Allows the user to supply starting values for the parameters
  used in refining the transformation between predicted and observed
  reflection positions.

2 XTOFRA
   XTOFRA <x>
   Crystal to detector distance expressed as a multiplier of the
   value given on DISTANCE keyword. 
   Default 1.0

2 YSCAL
   YSCAL <x>

   If the pixel sizes in slow and fast directions in the image are
    unequal, this is dealt with by the YSCAL parameter (which is
    refined). YSCAL is defined as the ratio of the pixel sizes in the
    slow and fast directions. Note that if only one pixel size is
    given (on a PIXEL keyword) it MUST be the
    pixel size in the "slow" direction in the image.

    Note that a value given here will overwrite the value calculated
    from the pixel sizes in X and Y if both were given on a previous
    PIXEL keyword.

   For R-AXIS image plate data the pixel sizes are different in
   the X and Y directions, and this is dealt with by YSCALE. (the
    correct value is 1.0324)

   Default 1.0, but 1.0324 for Raxis scanners

2 TILT
   TILT <x>
   Allows for a small rotation of the detector about a horizontal
   axis. Units are approximately 1/100ths of a degree.
   Default 0


2 TWIST
   TWIST <x>
   Allows for a small rotation of the detector about a horizontal
   axis. Units are approximately 1/100ths of a degree.
   Default 0

2 ROFF 
   ROFF <x>

   Radial offset (mm). This will allow for mechanical misalignment of
   Mar Research or Mac Science scanners in the scanning direction.
   Default 0.0

2 TOFF
   TOFF <x>
   Tangential offset (mm). This can correct for 
   misalignment resulting from the scan not going exactly through
   the centre of rotation in Mar Research or Mac Science scanners.
   Default 0.0

2 RDROFF
   RDROFF <x>

   Applicable to Mar and Dip scanners only.

   Radially dependent radial offset (mm). This allows for a non-linear
   scan on a Mar/Dip type scanner. The variation of the radial offset
   follows a sine wave, amplitude RDROFF mm, with NODE (see below)
   zero values between the centre and outer edge of the image plate.
   Default 0.0. By default this parameter is NOT refined

2 RDTOFF
   RDTOFF <x>

   Applicable to Mar and Dip scanners only.

   Radially dependent tangential offset (mm). This allows for a
   non-linear scan on a Mar/Dip type scanner. The variation of the
   tangential offset follows a sine wave, amplitude RDTOFF mm, with
   NODE (see below) zero values between the centre and outer edge of
   the image plate. 
   Default 0.0. By default this parameter is NOT refined

2 NODES
   NODES <n>

   Applicable to Mar and Dip scanners only.

   The number of zero values in the sine wave dependence of the
   radially dependent distortion parameters RDROFF, RDTOFF (qv). Only
   values greater than 1 are sensible.
   Default 2

1 DIVERGENCE
  DIVERGENCE <x [x]>
        Horizontal and vertical widths of the rocking curve (degrees).
   If only one number given then vertical divergence set to horizontal
   value. In this context, horizontal means the plane containing the
   rotation axis and X-ray beam.

   Default is 0 degrees for both.
 
1 DNAME
  DNAME <datasetname>

   Define a dataset name for use in data harvesting. Not yet fully
   implemented. MUST be used in conjunction with the PNAME keyword (qv).


1 DSTARMAX
  DSTARMAX <x>
        The maximum resolution in dimensionless r.l. units. (default set by
        maximum radius on detector).  The maximum resolution can also be set
        using the RESOLUTION card.  If neither DSTARMAX nor RESOLUTION is
        given then the maximum resolution defaults to that determined by
        maximum radius on the detector.
 
1 END
            PROGRAM STOPS

1 EXTENSION
   EXTENSION <odext>
   The extension for the image files.

   Default: "image" for Mar images, "osc" for R-axis, "ipf" for Mac
   Science scanners, "img" for FUJI, "corr" for prototype Mar images
   and "image" for unknown detectors.

1 FIDUCIALS
  FIDUCIALS 

   Sets search box sizes, threshold and positions of fiducials for
   processing images from offline scanners, eg Fuji, Molecular
   Dynamics. THIS IS NOT FULLY IMPLEMENTED AT PRESENT.

2 SEARCH
   SEARCH <XMMF>
   Search box size of XMMF mm for finding fiducials (NOTE the direct
   beam search box size is set by BEAM subkeyword).
   Default: Set in site specific code ?

2 THRESHOLD
   THRESHOLD <THRESH>
   Threshold used in fiducial search (counts)
   If the search is unsuccessfull, the threshold is automatically
   lowered to 5 sigma above background.

2 POSITION
   POSITION <FIDXY(NFID,2)>
   The coordinates (in mm) of the fiducials, x then y for the
   NFID fiducials (NOT including the direct beam).
   Default: Enraf Nonius fiducial coordinates.

2 BEAM
   BEAM <MMDB>
    Search box size for the direct beam (mm). If zero, no
   direct beam search is performed. This value should not be made 
   too large, or low resolution spots may fall in the search area.
   Default 0

3 More
    Initial camera constants CCX,CCY can be determined from  the
    position  of  the  direct beam (if present).  To invoke this
    option MMDB should be set to the  dimension  in  mm  of  the
    square  to  be  scanned  around  the  expected  direct  beam
    position.  This value should not be made too  large  or  the
    program  may  pick  up blackening caused by scatter from the
    beamstop.  If  no  direct  beam  present,  set  MMDB  to  0,
    otherwise a typical value is 2.5.

1 FINDSPOTS
   FINDSPOTS
   Find spots for autoindexing. There are several parameters which
   can be set for the spot search, each controlled by its own keyword.

2 FIND
  FIND <image number>

  Performs the spot finding; this can be used independently of 
  autoindexing. There is no reason to use this command under normal
  circumstances; it has been included for use in automated data 
  processing. An image TEMPLATE (or IDENT and EXTENSION) must also be 
  given.

2 THRESHOLD
  THRESHOLD <thresh>

   Sets the threshold for considering a pixel to be part of a
   spot. Pixels with values greater than <thresh>*sigma above the
   background at that radius are considered to be part of a spot,
   where the sigma is determined when finding the radial
   background.. This value should be increased to between 20 and 100
   for very strong images.

   Default: 5

2 NOICE
  NOICE

   Do not exclude spots that are in resolution shells that match those
   of ice rings.

   Default: Do exclude such spots

2 ICE
  ICE

   Do exclude spots that are in resolution shells that match those
   of ice rings.

   Default: This is the default, but if turned off by NOICE it can be
   turned on again by ICE

2 THRESHMAX
  THRESHMAX <maxthresh>

   If a value for the threshold is not given explicitly with the
   THRESHOLD keyword, the program will attempt to determine a suitable
   value (low for weak images, high for strong images). This value
   will be reset to "maxthresh" if it is calculated to be higher than
   "maxthresh".

   Default: 50

2 RMIN
  RMIN <rminsp>

   Sets the minimum radius <rminsp> (in mm) for the spot search,
   measured from the direct beam position.

   Default 0.1*Radius of image

2 RMAX
  RMAX <rmaxsp>

   Sets the maximum radius <rmaxsp> (in mm) for the spot search,
   measured from the direct beam position.

   Default 0.9*Radius of image

2 SPLIT
  SPLIT <xsplit> <ysplit>
   Spot splitting parameter in X and Y (mm). Any pixels above
   threshold which lie within xsplit,ysplit of the centroid of a
   previously found spot will be deemed to be part of that spot. This
   enables the algorithm to deal with split spots.

   Default 0.3 0.3 (mm)

2 MINX
  MINX <minxsize>

   Sets the minimum spot size in X as a function of the median spot
   size. 

   Default 0.5

2 MAXX
  MAXX <maxxsize>

   Sets the maximum spot size in X as a function of the median spot
   size. When finding spots on a very intense image with close spots
   it may be necessary to reduce <maxxsize> (eg to 2.0) to avoid two
   adjacent strong spots being treated as a single spot (this will
   happen if the counts do not drop to the threshold level between the
   two spots).

   Default 3.0

2 MINY
  MINY <minYsize>

   Sets the minimum spot size in Y as a function of the median spot
   size. 

   Default 0.5

2 MAXY
  MAXY <maxysize>

   Sets the maximum spot size in Y as a function of the median spot
   size. When finding spots on a very intense image with close spots
   it may be necessary to reduce <maxysize> (eg to 2.0) to avoid two
   adjacent strong spots being treated as a single spot (this will
   happen if the counts do not drop to the threshold level between the
   two spots).

   Default 3.0

2 MINPIX
  MINPIX <npixmin>

   Sets the minimum number of pixels in a spot.

   Default 6

2 XOFFSET
  XOFFSET <xoff>

   Sets the displacement (in mm) for the strip used to determine the
   radial background which will otherwise be centred on the direct
   beam position. The default direction for this strip is normal to
   the direction of the rotation axis (whether this is along X or Y),
   but if XOFFSET is specified then the background strip will be
   placed parallel to the Y axis.

   Default: Zero displacement. Orient the background strip at right
   angles to the rotation axis (to avoid any possible backstop shadow
   on the image).


2 YOFFSET
  YOFFSET <yoff>

   Sets the displacement (in mm) for the strip used to determine the
   radial background which will otherwise be centred on the direct
   beam position. The default direction for this strip is normal to
   the direction of the rotation axis (whether this is along X or Y),
   but if YOFFSET is specified then the background strip will be
   placed parallel to the X axis.

   Default: Zero displacement. Orient the background strip at right
   angles to the rotation axis (to avoid any possible backstop shadow
   on the image).

2 XMIN
  XMIN <x>

   Spots with an X coordinate (relative to the direct beam position)
   of less then "x" mm will be rejected.

   Default 0.0

2 YMIN
  YMIN <y>

   Spots with an Y coordinate (relative to the direct beam position)
   of less then "y" mm will be rejected.

   Default 0.0

2 RADIAL

   Selects (old) radial background determination. (Default)

2 LOCAL

   Selects local background determination. This can be useful for weak
   images or those with asymmetric background.

2 BOX
  BOX <x> <y>

   The local background will be determined within boxes of size "x" by
   "y" pixels.

   Default 100,100

2 MAXPEAKS
  MAXPEAKS <n>

   Any spot with more than "n" local maxima will be rejected

   Default 20

2 MAXYEXTENT
  MAXYEXTENT <maxy>

   Maximum size of a spot (in pixels) in the fast changing (Y)
   direction in the image.

   Default 50

2 RMSFAC
  RMSFAC <x>

   For spots containing multiple peaks, the rms variation over the
   pixels in the peak must be more than "x" times the rms variation
   expected from counting statistics if all pixels had the threshold
   value.

   Default 3.0

1 GAIN
   GAIN <x>

   For image plate detectors, this specifies the gain of the detector
   in pixel values per absorbed X-ray photon.  It is CRUCIAL to have a
   reasonable estimate of the gain as many aspects of the program use
   counting statistics derived standard deviations to determine
   acceptable spots for refinement, profiles etc. The correct value of
   the gain should give a BGRATIO of unity, and the actual value of
   BGRATIO obtained can be used to get an improved estimate of the
   gain using the relationship:

    true gain  = estimated gain * (bgratio)**2

   See the User Guide for a way of estimating the GAIN.

    Defaults:
P
    Detector			Default gain
    Mar				1.0
    Mar CCD			1.0
    ADSC Quantum 4 CCD		0.5
    Raxis II			5.0
    Raxis IV			1.0
    CHESS 1K			0.5
    CHESS 2K			0.5
    CHESS ADSC (1 module)	1.0
    DIP2000			1.0
p
1 GENFILE
  GENFILE <genfilename>
   The name of the generate file  (up  to  200  characters)  The
   suffix .gen is automatically added if not specified. The generate
   file is used as a scratch file during processing. Note that if two
   jobs are running from the same directory, they must have unique
   generate filenames.

   If no generate filename is given, the default is the MTZ output
   filename but with the suffix ".gen"

1 GO
  GO
  Identical to the RUN keyword (see elsewhere).

1 HARVEST

  Sets up data harvesting. At some stage it is proposed that this will
  become compulsory, but at present it is only included for testing
  purposes.

2 PNAME   <identifier>    (compulsory)

    This is a user supplied identifier that will be used throughout
  the structure solution to describe the project under study, either a
  protein_name or nucleic acid name that will correspond when
  published and deposited to the PDB to an entry/IDcode. The
  identifier should be suitable for a file name, i.e. not contain any
  white space nor contain characters that would be disallowed in the
  creation of a unix or vms file name.

2 DNAME  <identifier>  (compulsory) 

    This is a user supplied identifier that will be used throughoutthe
  structure solution to describe each particular dataset that is
  collected to solve any particular Project name.

    NOTE: in the case of a protein, the datasets for
P
       Native
       mutant
       native_complex
       mutant_complex 
p
  would each be finally published/deposited as unique PDB entries and
  these correspond to PNAME. DNAME refers to for example each dataset
  collected for a Native and heavy atom derivatives or for each
  wavelength for a MAD data set.

2 Overview

   For both PNAME and DNAME these names in the context of harvesting
  for deposition (see http://www2.ebi.ac.uk/msd/Harvest/index.shtml)
  these items are planned to be compulsory for all programs that can
  produce information for deposition. However, in the case of CCP4
  programs that read MTZ data files, these items will be got from the
  mtz header, and should only be required as keyworded input for those
  programs that are used as entry points to creating a first MTZ file,
  such as mosflm, f2mtz and rotaprep.

    One PNAME and DNAME will together describe a set of coordinates
  and the dataset the refined coordinates refer to upon final
  deposition.
P
    In mmCIF terms PNAME corresponds to _entry.id
    and            DNAME corresponds to _diffrn.id
p
  examples:

   (i) example where one knows one is collecting just 1 dataset and
   using both PNAME and DNAME to describe a complex
P
   PNAME    phosphate_binding_protein
   DNAME    A197C_chromophore
p
   one could also use
P
   PNAME    phosphate_binding_protein_A197C_chromophore
   DNAME    native_complex
p
   (ii) example where one will later use MIR to solev the structure
P
   PNAME Xylose_Isomerase
   DNAME Native
p
    (iii) example of using the new keywords
P
    PNAME rac1
    DNAME complex_GNP_Mg
    ! TKDC is temp of data collection in degrees Kelvin
    ! PHDC is PH of data collection
    ! CRYSTALLISATION is any string of words describing crystallisation details
    XDETAILS TKDC 101 PHDC 6.8 CRYSTALISATION -
     '20% PEG 8k, 4oC, 20mM TrisHCL buffer pH 7.6, 10mM MgCl2, 5mM DDT'
p
2 UCWD 

    If this keyword is supplied then the harvest mosflm snapshot file
  will be written to the current working directory. This will be
  automatically set .TRUE. if the $USER is recognised as either
  'raxis' or as 'ipuser', where the user of mosflm does not have a
  $HOME directory (or SYS$LOGIN directory) that is unique to
  themselves. In this case the user then will have to move the
  harvest/deposition file to the Deposit directory at a later date.
  This is also useful when running tests on image data processing and
  one is not certain that the particular run will be the final run of
  mosflm. When UCWD (use_current_working_directory) is set .TRUE. the snapshot
  file is created as:
P
   DNAME.mosflm_run_X_Y
p
  where X is the first image number and Y is the last image number as
  found in the summary file type of output, therefore if more than one
  RUN keyword is used with different PROCESS keywords the overall
  range of images processed is got from the internal summary information.

   when UCWD is set .FALSE. then the same file is created in the directory
   $HOME/DepositFiles/PNAME/


2 XDETAILS (optional) 
   The sub-keywords for XDETAILS supplies information for deposition
   that should be known at the time of data collection but the information
   is not used by mosflm, this currently are,
   subkeywords
P
     TKDC    The temperature of data collection in degrees kelvin
     PHDC    The pH of the crystal used in data collection
     CRYSTALLISATION   'string'  a quoted string to describe
                       details of the crystallisation
     e.g. '20% PEG 8k, 4oC, 20mM TrisHCL buffer pH 7.6, 10mM MgCl2, 5mM DDT'
p


-----------------------------------------------------------------------





1 HKLOUT
  HKLOUT <mtzfilename> [MULTIPLE | NOMULTIPLE]

   The name of the MTZ file (up to 200 characters) The suffix .mtz is
   automatically added if not specified. Note that if two jobs are
   running from the same directory, they must have unique MTZ
   filenames.

   If the subkeyword MULTIPLE is given, a new mtz file will be created
   for each "block" of images that are processed. This allows data to
   be fully processed (ie run through SCALA) while it is still being
   integrated/collected (see the WAIT keword also). NOMULTIPLE turns
   off this mode. In the case of using multiple MTZ files, the MTZ
   filename supplied by the HKLOUT keyword is modified by appending
   "_001" for the first file, "_002" for the second etc.
P
   eg. HKLOUT myog3x1.mtz MULTIPLE
p
   will generate MTZ filenames:
P
   myog3x1_001.mtz                for the first block of images
   myog3x1_002.mtz                for the second block of images etc
p
   **** If the MTZ filename is assigned in the command line, and the
   HKLOUT keyword is just used to trun on the MULTIPLE MTZ file
   option, the the entire subkeyword (MULTIPLE) must be given, it
   cannot be abbreviated ***

   Defaults:
   If no MTZ filename is given, a default filename made up of the
   image identifier and the first image number is created. 

   A single MTZ file is used for all images processed (ie NOMULTIPLE)

   Note that the name of the MTZ file can also be given in the command
   line with HKLOUT.

1 IDENTIFIER
  IDENTIFIER <template>

        The template (up to 40 characters in length) is used to make
  up the image filenames, which are of the form "template_###.extension"
  where "extension" is defined by EXTENSION which defaults to "image"
  for Mar images, "osc" for R-axis.

  For a more general way to define image filenames see keyword TEMPLATE.
 
1 IMAGE
  IMAGE <imagefilename|image number>  [PHI <phistart> [TO] <phiend>]

   This is the best way to start an interactive graphics session.
   It allows an image to be inspected using the X-windows graphics
   without needing to define any crystal or detector parameters. It is
   useful for checking the quality of a diffraction pattern prior to
   autoindexing or processing, or for locating the direct beam
   position. The full filename of the image file should be given. For
   Mar scanners this is the only keyword required (other than the RUN
   or GO keyword), for other scanners the scanner type must be defined
   using the DETECTOR keyword.
P
   eg IMAGE /ss2/andrew/f1/images/f1x32_001.pck
      RUN

   or IMAGE /ss2/andrew/f1/images/f1x32_001.pck
      DETECTOR RAXISIV
      RUN
p

   NAMING CONVENTION FOR IMAGES
   ============================

   It is assumed that the images conform to a naming convention where
   the image name is made up of three parts, a template, a three digit
   number and an extension. The template can be up to 40 characters
   long, and should be separated from the three digit number by a
   hyphen (-) or an underscore (_). The extension can be up to 8
   characters long and should be separated from the three digit number
   by a period (.). Note that the template can contain underscores or
   hyphens.

   Examples of valid image filenames are:
P
   lysozyme_cryst1_021.image
   catx1_001.img
   f1_tray42_wellb6_001.osc
p
   ALTERNATIVE NAMING CONVENTION
   =============================

   If the image filenames do not conform to the specification given
   above, the TEMPLATE keyword can be used to define a very general
   format for the image filenames. 
   If the TEMPLATE keyword is to be used, it MUST preceed the IMAGE
   keyword in the input, and the image NUMBER, not the filename,
   should be given. See TEMPLATE for more information.
P
   eg TEMPLATE fred_###
      IMAGE 23 PHI 22 TO 23
p
   will read the file "fred_023".
       
1 LATTICE

   LATTICE RHOMBOHEDRAL

    By default, rhombohedral space groups are processed using the
    equivalent hexagonal setting. If the rhombohedral cell is to be
    used rather than the hexagonal cell, the LATTICE RHOMBOHEDRAL
    keyword must be given. This keyword MUST be given BEFORE the
    SYMMETRY keyword. The symmetry operators for the rhombohedral
    setting are not included the the "symop.lib" library supplied with
    the CCP4 distribution. They must be added, and MOSFLM requires
    that the space group number is given on the SYMM keyword, and that
    this number is the true spacegroup number plus 1000. Thus R3
    should be specified as: SYMM 1146 and R32 as: SYMM 1155.
    The entries to be added to "symop.lib" for these two spacegroups
    are given below.

    Example input to MOSFLM
P
    TITLE Testing  rhombohedral cell
    CELL 74.5 74.5 74.5 92.3 92.3 92.3
    CRYST R
    SYMM 1155
    AUTOINDEX
    PROCESS 1 TO 20
    GO
p


1 LIMITS
  LIMITS XMIN <x> XMAX <x> YMIN <x> YMAX <x> RMIN <x> RMAX <x>
  LIMITS XSCAN<x> YSCAN <x> RSCAN <x> RSCAN <x> CENTRE <x y> RCENTRE <x y>
  LIMITS EXCLUDE <x1 y1 x2 y2> or LIMITS RECTANGLE <x1 y1 x2 y2>
  LIMITS QUADRILATERAL <x1 y1 x2 y2 x3 y3 x4 y4>
  LIMITS ARC <x y radius phihigh philow INSIDE/OUTSIDE>

        Defines usable area of detector.  Several key words can be
        repeated on one line or several limits cards can be specified.
        Note that XMIN,XMAX,YMIN,YMAX,RMIN,RMAX are all with respect
        to the direct beam position. RSCAN,XSCAN and YSCAN however are
        with respect to the central pixel in the image (ie 600,600 for
        the small (90cm) Mar scanners and 1000,1000 for the large Mar) but
        this can be changed using subkeyword CENTRE (below). X and Y
        are in the detector coordinate frame, ie with X along the slow
        changing direction in the image, Y in the fast direction.

        XSCAN and YSCAN are ONLY used for "RECTANGULAR" detectors
        (such as the Raxis and CCD detectors) while RSCAN is only used
        for "CIRCULAR" detectors such as the Mar and Mac Science.

        RCENTRE and RMIN can be used to define the position and size
        of the backstop shadow, but this is usually done with thE
        BACKSTOP keyword (qv).

        Possible keywords are:
P
                                                Defaults (mm)
                                                  IP          	
                                        180mm Mar  300mm Mar  RAXIS
        XMIN x          minimum X         0           0        	0
        XMAX x          maximum X        90          150      100
        YMIN x          minimum Y         0           0        	0
        YMAX x          maximum Y        90          150      100
        RMIN x          minimum radius      0.05*MIN(XMAX,YMAX)
        RMAX x          maximum radius   90          150      141.4
p
        ** all the above are with respect to the direct beam position **


        IMPORTANT: For the XMIN,XMAX,YMIN,YMAX tests, the ABSOLUTE
        VALUE of the spot coordinate on the detector relative to the
        direct beam position is checked. Thus these limits must always be
        given as POSITIVE values or they will have no effect. These
        limits will always be symmetrical relative to the direct beam
        position. Use the EXCLUDE keyword (below) to define limits
        that are not symmetric. 
P
        XSCAN x         Maximum distance                      100
                        from centre of
                        detector in X
                        (slow direction
                        in image)                
        YSCAN x         Maximum distance                      100
                        from centre of
                        detector in Y
                        (fast direction
                        in image)                
        RSCAN x         maximum radius   90          150       -
                        from center of
                        detector
        CENTRE x y      defines the centre of the circle from which RSCAN
                        is measured. x and y should be given in mm 
                        relative to an origin at the first pixel in the
                        image (as for the direct beam coordinates).
                        x,y default to the centre of the image.

        RCENTRE x y     defines the centre of the circle from which RMIN
                        is measured. x and y should be given in mm 
                        relative to an origin at the first pixel in the
                        image (as for the direct beam coordinates).
                        x,y default to the direct beam
                        coordinates. Use the BACKSTOP keyword instead.


                + = MIN(SQRT(XMAX**2+YMAX**2),58)
p
        Note that X is the slowly varying direction in the image. In
        Mar and R-axis scanners this is parallel to the rotation axis.
        Y is the fast varying direction in the image.

        If not specified, these values (except RMIN) will default to
        the appropriate physical size of the detectors. Note that
        giving a value of zero WILL NOT WORK, it will simply revert
        to the default.
P
        EXCLUDE <x1 y1 x2 y2> or RECTANGLE <x1 y1 x2 y2>
p
        This enables a rectangular area with opposite corners defined
        by the coordinates x1,y1 and x2,y2 to be excluded from spot
        finding and measurement. The coordinates are with respect to
        an origin at the first pixel in the image, and with X along
        the slowly varying direction in the image. This is the same
        coordinate frame as that used to define the direct beam
        coordinates. 

	Up to 10 rectangular areas can be excluded.
P
	QUADRILATERAL <x1 y1 x2 y2 x3 y3 x4 y4>
p
	This enables a convex quadrilateral with corners defined by
	(x1,y1) etc. to be excluded from spot finding and measurement.
	The coordinates are as defined for the EXCLUDE sub-keyword.

	Up to 10 quadrilaterals can be excluded.
P
	ARC <x y radius phihigh philow INSIDE/OUTSIDE>
p
	This enables regions defined by arcs to be excluded from spot finding 
	and measurement. Either an internal (i.e. sector or ice-cream cone shaped) 
	or external region can be excluded.
	
	The coordinates are as defined for the EXCLUDE sub-keyword; (x,y) defines 
	the centre of the arc; phihigh and philow the angles of the radii in 
	degrees.

	Up to 10 arcs can be excluded.


1 MATRIX 
  MATRIX <x x x x x x x x x> OR <filename>
        The A matrix as A11,A12 etc. specifying the crystal
        orientation and cell parameters  Not required if U matrix given.
        If a filename is given here, then the [A] matrix AND the
        missetting angles will be read from that file, format (3F12.8).

1 MAXWIDTH
  MAXWIDTH <x>

        This sets the maximum width for a summed partial in degrees.
        The program converts this to the equivalent number of images.
        There is an upper limit of 100 images.
        Default: 5 degrees
 
1 MISSET
  MISSET <x x x> (RESET)

        The missetting angles (degrees). If this keyword is given,
        thes values will OVERIDE values read from a file using the
        "MATRIX" or "UMATRIX" keywords.
        If the subkeyword RESET is given, then the missetting angles
        will be incorporated into the A (or U) matrix and the actual
        misseting angles will be reset to zero. This can be useful
        when adjusting the orientation manually.

        Default: zero, or values read from MATRIX or UMATRIX file.

1 MOSAIC
  MOSAIC <x> or <ESTIMATE>

        Mosaic spread of crystal in degrees.
        Defaults to zero if not specified. For standard source it is
        normally included in divergence (DIVERGENCE keyword).

        MOSAIC ESTIMATE returns a value estimated from an analysis of 
        integrating the current image (or first image in a list used 
        for autoindexing) with a range of different mosaicities; it 
        may only be used in a run with AUTOINDEX DPS.

        Note that the mosaicity may well increase during data
        collection due to radiation damage. This should be monitored
        using postrefinement and the mosaicity either increased
        accordingly or the refined value used in processing (POSTREF
        USEBEAM.)

1 NEWMAT 
  NEWMAT <filename >
        
        When doing AUTOINDEXING or POSTREFINEMENT SEGMENT (qv) option
        to determine accurate cell parameters, the final orientation
        matrix (the A matrix) will be written to "filename", in a
        suitable format to be input to a processing job using the
        MATRIX filename keyword.

        Default filename: NEWMAT

1 NOLP
  NOLP
         Do NOT apply the Lorentz Polarisatio correction to the
        intensities written to the output MTZ file.

        Default: Apply Lp correction


1 NOMEASURE
  NOMEASURE

      If this keyword is present, the images will not actually be
   integrated. However, refinement of detector parameters and
   post-refinement will be performed. This is only for special
   applications. 



1 NULLPIX
  NULLPIX <x>

   Pixels with a value less than or equal to "x" will be treated as
   being outside the active area of the detector. Any spots containing
   pixels with values les than or equal to "x" this value will be
   rejected (unless using PROFILE EDGE option).

   This provides a way of dealing with "shadows" on the detector.

   Default: 0 for all detector except CHESS CCD where value is 100.

1 OVERLOAD

   OVERLOAD NOVER <x> CUTOFF <x>
   Overloaded reflections are those with more than a preset number
   of pixels with values greater than a supplied threshold.
   Overloads are normally excluded from refinement, always
   excluded from profiles, and normally flagged as badspots.
   (However their intensities can be estimated using profile
   fitting, see PROFILES keyword.) 

2 NOVER
   NOVER <NOVPIX>
   Reflections with more than NOVPIX pixels with values greater
   than the cutoff are flagged as overloads.
   Default: 0

2 CUTOFF
   CUTOFF <x>
   Sets the threshold for overloads. (Note that the threshold
   used in forming the standard profiles is set using CUTOFF
   subkeyword on the PROFILE keyword).
   Defaults:
P
   Commercial Mar scanners: 100000
   Prototype EMBL scanners:  31999
   Chess Fuji Scanner:        9999
   Raxis II:                250000
   MacScience Dip2000       500000
p
1 PIXEL
  PIXEL <x [y]>

     The image pixel size (mm) in the "slow" direction in the image.

    If a second value is given, this is taken as the pixel size in the
    "fast" direction and used to calculate the YSCAL parameter, which
    defines the relative pixel sizes in X and Y . The value calculated
    here (as x/y) will overwrite any value given by previous
    DISTORTION YSCAL keywords.

    The pixel size in the "fast" direction is defined to be x/YSCALE
    and initial values of YSCALE can also be specified on the
    DISTORTION keyword. Eg. For the R-axis, YSCALE is 1.032).

    Default: Depends on detector type, set automatically by program.

1 PLOT 
     PLOT [ <N1 [N2]> ]

   Display the image using X-windows graphics with the calculated
   pattern overlaid. The cell parameters, orientation, mosaic spread
   etc can then be adjusted to improve the fit between the calculated
   pattern and the image.


1 PNAME
  PNAME <projectname>

   Define a project name for use in data harvesting. Not yet fully
   implemented. Should be used in conjunction with the DNAME keyword (qv).

1 POLARISATION
  POLARISATION [PINHOLE MIRRORS MONOCHROMATOR SYNCHROTRON <x>]

  Defines the polarisation of the X-ray beam. PINHOLE and MIRRORS both
  define an unpolarised beam, MONOCHROMATOR defines the polarisation
  for a GRAPHITE monochromator, and for SYNCHROTRON sources the
  polarisation can be defined using the SYNCHROTRON subkeyword.

  Default: An unpolarised beam is assumed. If the SYNCHROTRON
  subkeyword is given, the default polarisation is 0.86 (appropriate
  for the SRS)


1 POSTREF

   This invokes post-refinement to refine missetting angles and cell
   parameters and beam parameters.  It can ONLY be called where it is
   appropriate to sum partials over abutting images (ie successive
   images have been recorded with the same exposure time). The program
   will detect cases where adjacent images have apparent exposure
   times that differ by more than 5% and issue a warning message.By
   default, postrefinement is used (POSTREF OFF will suppress
   postrefinement) and for crystals of trigonal or higher symmetry it
   will use data from a single pair of images and will refine cell and
   orientation, while for orthorhombic or lower symmetry it will use
   10 degrees of data (or any other wedge as specified by the user) to
   refine cell and orientation.

   The POSTREF SEGMENT option (see below) is a convenient way of
   determing accurate cell parameters using data in a number of
   different segments over a wide phi range.

   The way the postrefinement is implemented now depends on whether
   the detector is considered to be ONLINE or OFFLINE (see DETECTOR
   keyword). However in works equally well in both cases.

   The POSTREF MULTI option is now the default, and is suitable for 
   processing images where the sum of the mosaic spread and beam
   divergence is more than twice the oscillation angle. In some very 
   rare cases, it may be worthwhile using the old-style postrefinement
   by employing the POSTREF NOMULTI option.

2 More

   Following positional refinement, the measurement boxes for all
   reflections on the image are extracted from the image. THese are
   used for post-refinement in a way that depends on whether the
   detector is ONLINE (as Mar and Raxis are by default) or OFFLINE (as
   MD is by default).

   Post refinement when mosaicity is greater than 2 oscillation widths.
  =====================================================================

   The default mode of post refinement will not work if the combined
   mosaic spread and beam diveregence is greater than twice the
   oscillation angle. To process images in these circumstances, use
   the POSTREF MULTI option.


   Post refinement for ONLINE detectors
  =====================================

   When summing partials, two measurement boxes are obtained for those
   reflections which are partially recorded at the end of the
   oscillation range. The first corresponds to the pixel values for
   that reflection in the current image, the second corresponds to the
   sum of the pixel values for the current image and the following
   image (ie the fully recorded reflection). Both these boxes are
   integrated, giving the integrated intensity for the partial and the
   fully recorded reflection. These intensities are then used to
   refine crystal orientation and optionally cell parameters and beam
   parameters.

   Post refinement for OFFLINE detectors
   ====================================

   Partially recorded reflections at the end of the current
   oscillation are stored in memory for each image. Partially recorded
   reflections at the start of each image are matched up with the
   corresponding reflection on the previous image (stored in memory)
   to give the intensities of the two halves of the partial.  These
   intensities are then used to refine crystal orientation and
   optionally cell parameters and beam parameters.


   Different types of Post refinement
   ==================================


   There are three different modes of operation, depending on how much
   data is included in the post-refinement.

   1) If data from only a single pair of images are used (SINGLE),
   then the crystal orientation around the X-ray beam is very poorly
   determined and will not be refined. Missetting angles PSIY, PSIZ
   corresponding to rotations around camera Y and Z axes at the
   current phi value are refined (note that PHIX, PHIY,PHIZ are
   rotations applied at phi=0), and these are converted to the
   equivalent PHIX,PHIY,PHIZ missetting angles after refinement.  It
   should be realised that for crystal symmetry lower than trigonal
   (ie more than 2 cell parameters) the cell is not well defined when
   using data from a single pair of abutting images. Thus in this mode
   cell refinement is NOT recommended for symmetry lower than trigonal
   (use multiple images as described below). Even with trigonal and
   higher symmetry, there may not be sufficient data (ie number of
   reflections) to refine the cell, particularly when the X-ray beam
   has very low divergence (as at a synchrotron source), resulting in
   rather few partially recorded reflections. The consistency of cell
   parameters and their estimated standard deviations should be used
   to gauge whether cell refinement using a single pair of images is
   appropriate (see below for alternatives). In particular, when cell
   refinement results in the missetting angles becoming poorly
   determined (ie sd's above 0.02) it should not be used.  Any
   combination of cell parameters can be FIXED (see below)

   This is the default mode of post-refinement for crystals of
   trigonal or higher symmetry.

   2) If data from a series of pairs of images are used (NADD or
   WIDTH) then all cell parameters will be refined by default.  (Use
   the FIX keyword to fix individual cell parameters) There are two
   possible modes of operation. In the default mode all the images
   included in the refinement have their own independent missetting
   angles, and hence as the orientation around the X-ray beam is not
   determined in this case only the missetting angles PSIY,PSIZ are
   refined (as in the SINGLE case). This means that post-refinement
   can be used to refine cell parameters even if the crystal is
   slipping.  Every reflection will contribute to the refinement of
   the cell (and beam) parameters, but will only contribute to the
   refinement of one set of missetting angles.

   Alternatively, if the FIXED keyword is included, all images included
   in the refinement will be forced to have the same missetting angles.
   This may be helpful if there are not enough partials on each image to
   allow a stable refinement, but if the crystal has slipped then it will
   not work properly.

   Post refinement of cell parameters will be delayed until the
   positional refinement has been carried out on the required number
   of images (NADD) (but the crystal orientation will be refined after
   every image as in the SINGLE case).  Post-refinement will then be
   carried out following positional refinement of every image, with
   the data from the latest image replacing that from the image with
   the smallest phi value currently included (ie a circular buffer
   with a fixed angular width).

   The stability of the refinement will depend on the resolution
   and strength of the data and the angular range included in the
   refinement.

   Typically a wedge of data of between 5 and 10 degrees gives
   a satisfactory refinement. The standard deviations of the
   refined parameters should be carefully checked to see that
   they are reasonably small. Even with a wedge of data, the cell
   parameter(s) closest to the X-ray beam may not be well determined.
   However, this parameter will also have least effect on the predicted
   spot positions and degree of partiality. It should also be realised
   that there may be significant correlations between different
   refined parameters, particularly in lower symmetry space groups
   (eg in  monoclinic space groups, rotating around the unique axis,
   there is a correlation between the missetting angle PHIZ, the beta
   angle and the relative a and c cell dimensions.) Thus individual
   refined cell parameters may well deviate significantly from their
   correct values, but this will not necessarily hinder the
   satisfactory processing of the images.

   This is the default mode of post-refinement for crystals of
   orthorhombic and lower symmetry.

   3) Using the SEGMENT mode, data for post-refinement is accumulated
   over a number of different "blocks" or "segments" of data (at
   different phi values) in order build up sufficient data that all
   cell parameters are well defined even in low symmetries. The data
   should all come from one crystal, or from crystals with a similar
   orientation (within 20 degrees of each other). This is the
   preferred procedure for orthorhombic or lower crystal symmetry.
   Data from segments spanning as wide a phi range as possible should
   be used. See User Guide for an example command file.


2 MULTI

   DEFAULT. Uses partials spread over 2 or more images (up to 20) in
   postrefinement. Can be used to allow processing of fine phi-sliced
   data, but should not be used for normal coarse sliced data (where
   the oscillation angle .GE. 0.5*mosaicity).

2 NOMULTI

   Turns off the POSTREF MULTI mode.

2 SEGMENT
  SEGMENT <NSEG>

   Use data from NSEG different segments in phi. Each segment is
   defined by a PROCESS keyword followed by a RUN keyword. Different
   orientation matrices (MATRIX keyword) and crystal identifiers
   (IDENT) can be given for each segment. If different orientation
   matrices are to be given for different segments, it is important
   that they are all with respect to the same phi origin.
P
   eg POSTREF SEGMENT 3
      PROCESS 1 TO 3 ANGLE 1.0 START 0.0
      RUN
      PROCESS 45 TO 47 ANGLE 1.0 START 44.0
      RUN
      PROCESS 88 TO 90 ANGLE 1.0 START87.0
      RUN
p
      ** NOTE ** Normally it is not necessary to give the oscillation
   angles (ANGLE) or starting phi values (START) as these are read from
   the image headers.

2 FIX
  FIX <keywords>

  Normally all variable cell parameters are refined when using 
  data from a series of images (NADD or WIDTH). To fix a given
  cell parameter, it must be "FIXED".
  Possible keywords are:
P
              'A' 'B' 'C' 'ALPHA' 'BETA' 'GAMMA' or 'ALL'

  eg POSTREF FIX ALL
p
  will fix all cell parameters, only orientation angles will be
  refined.

  Default: All variable cell parameters refined when using data
  from a series of images.

  From version 6.2.3 onwards, by default the refined mosaic spread
  will be used during integration. To force the program to use the
  input value rather than the refined one, use keywords: 
P
  POSTREF FIX MOSAIC
p

2 UNFIX
   UNFIX <keywords>

   Normally cell parameters are NOT refined when using data from a
   single pair of images (SINGLE) if the symmetry is lower than
   trigonal. To refine a given cell parameter, it must be "UNFIXED".
   Possible keywords are: 'A' 'B' 'C' 'ALPHA' 'BETA' 'GAMMA'

   Default: No cell parameter refinement when using a single pair
   of images for orthorhombic and lower symmetries for trigonal and
   higher symmetry all variable cell parameters are refined.

2 MOSSMOOTH <nsmooth>

   The refined mosaic spread will be smoothed over "nsmooth" images,
   to provide a more stable refinement.

   Default: 10

2 MOSADD <etafrac>

   A safety factor of "etafrac"*mosaic_spread will be added to the
   refined value of the mosaic spread to ensure the true mosaic spread
   is not underestimated.

   Default: 0.05

2 OFF
   This turns off post refinement

2 ADD
  ADD <N>

  Use data from N successive pairs of images for post-refinement.
  In this case, all three missetting angles and all variable cell
  parameters will be refined (see FIX to prevent cell parameter
  refinement). Missetting angles are refined independently for each
  image (unless subkeyword FIXED is used...see elsewhere). 

  This is the default mode of post-refinement for orthorhombic or
  lower symmetry with NADD calculated to give an angular wedge of data
  of 10 degrees.


2 WIDTH

  WIDTH <x>

  Use data from N successive pairs of images for post-refinement
  where N is calculated so that the resulting wedge of data covers
  an angular range of "x" degrees.
  In this case, all three missetting angles and all variable cell
  parameters will be refined (see FIX to prevent cell parameter
  refinement). Missetting angles are refined independently for each
  image (unless keyword FIXED is used...see elsewhere). 

  This is the default mode of post-refinement for orthorhombic or
  lower symmetry.

  Default: 10 degrees.

2 FIXED

   The missetting angles of all NADD images used in the postrefinement
   will be constrained to be the same. (See above)
   DEFAULT: Allow refinement of missetting angles for each image
   independently.

2 SINGLE

    Do post-refinement for every image using data from the current
   image and the next image only. For trigonal or higher symmetry cell
   parameters are refined unless FIXED explicitly, for lower symmetry
   they are not refined unless explicitly UNFIXED. Only two missetting
   angles PSIY,PSIZ are refined as the orientation around the beam is
   not determined (but refinement of CCOMEGA will allow for any
   rotation around the X-ray beam).

  (See More for further information). Large shifts (MAXSHIFT) 
  in orientation will result in the current image being remeasured.
  
  This is the default mode for trigonal or higher symmetry

2 SDFAC
  SDFAC <x>
  Only reflections for which the intensity is greater than "x"
  times the standard deviation will be used in postrefinement.
  Default 3.0           

2 NREF
  NREF <n>

  Post-refinement will ONLY be carried out if there are at least "n"
  reflections that can be used in the post-refinement. This is to
  prevent the refinement being carried out with very few reflections,
  as the resulting parameters will not be well defined.

  Default: 10

2 RESOL 
  RESOL <x1 [x2]>
  Only reflections out to a resolution limit of x1 angstroms,
  or if x2 is given, between x1 and x2 angstroms (may be given
  in any order), will be used for postrefinement. Suitable
  values for 2-3.5A data are 6A and 4A.
  Default: x1=1A (ie all possible reflections)

2 MAXRESIDUAL

  MAXRESIDUAL <x>
  If the angular residual following postrefinement is greater
  than x*(mosaic spread + beam divergence), processing will be abandoned.

  Default 1.0.


2 MAXSHIFT
  MAXSHIFT <x>
  a) SINGLE image refinement
  If the total shift (defined as the rms value of the sums of
  the shifts in PHIX,PHIY,PHIZ) is greater than x degrees, 
  the reflection list for the current image will be regenerated
  using the refined parameters and the image will be reprocessed.

  b) MULTIPLE image refinement (NADD or WIDTH)
  If the total shift (defined as the rms value of the sums of
  the shifts in PHIX,PHIY,PHIZ) is greater than x degrees, 
  for the FIRST post-refinement, remeasure the first NADD images
  again and repeat the post-refinement. This is done up to
  a maximum of NRPT times. For subsequent refinements, a warning
  is printed on the summary file, but there is no repeat processing.

  Default 0.05 degrees. Should be set larger for crystals with a
  high mosaic spread (eg 0.1 degrees)

2 SHIFTFAC
  SHIFTFAC <x>
  Only applicable to post-refinement using several images (NADD or
  WIDTH). If the shift in any refined cell parameter is 
  greater than x times its estimated standard deviation
  for the FIRST post-refinement, remeasure the first NADD images
  again and repeat the post-refinement. This is done up to
  a maximum of NRPT times. For subsequent refinements, a warning
  is printed on the summary file, but there is no repeat processing.

  Default 2.5

2 REPEAT 
  REPEAT <NRPT>
  Only applicable to post-refinement using several images (NADD or
  WIDTH).  Maximum number of repeats allowed if there are large
  shifts in orientation (MAXSHIFT) or cell parameters (SHIFTFAC).

  Default 5

           

2 BEAM
   BEAM <x>
   Sets refinement of beam parameters:
P
     x = 0   switches off refinement
p
    For convential source:
P
     x = 1   isotropic divergence refined
     x = 2   both horizontal and vertical divergence
      are refined
p
    For synchrotron source:
P
     x = 1   only mosaic spread refined
     x = 2   mosaic spread not refined, horizontal
      and vertical beam divergences refined
p
2 USEBEAM [OFF]
  USEBEAM
  If this keyword is present, the refined beam parameters will be
  used in generating the reflection list.

  Default: do not use refined beam parameters

2 FRMIN
  FRMIN <fracmin>

  Only reflections which are more than a fraction "frmin" recorded
  will be used in post-refinement.
  default 0.05

2 FRMAX
  FRMIN <fracmax>

  Only reflections which are less than a fraction "frmax" recorded
  will be used in post-refinement.
  default 0.95

1 PRINT 
     PRINT [options] 

   Allows more information to be printed on the image processing.  The
   options allowed are 'FIDUCIALS', 'REFINEMENT', and 'STATISTICS'.  The
   FIDUCIALS option gives additional information on the location of the
   fiducial marks (if any). The REFINEMENT option gives additional
   information on refinement of detector parameters using spot positions (eg
   list of rejected reflections, residuals in initial cycles etc). The
   'STATISTICS' option gives a breakdown of the agreement of profile fitted
   and integrated intensities as a function of both intensity and box
   number, for fully recorded and partials separately.  This keyword should
   normally be present for the first few images processed from any crystal,
   but can then safely be omitted.  If no 'options' are given, ALL options
   are turned on.  

   DEFAULT: When running interactively, FIDUCIAL and REFINEMENT options are
   turned on., when in background, all options are off.

1 PROCESS

  PROCESS <img1 [TO] img2> [START <phist> [ANGLE/OSC <oscrange>]]
                      [BLOCK  <nblock>  ADD <iseradd>]
P
  eg PROCESS 1 TO 35 START 25.0 ANGLE 0.5 BLOCK 6 ADD 1000
  or PROCESS 1 35 START 25.0 ANGLE 0.5 BLOCK 6 ADD 1000
  or PROCESS 1 TO 35 START 25.0 OSC 0.5 BLOCK 6 ADD 1000
p  
  This keyword specifies which images are to be processed. It replaces
  the SERIAL keyword which is now obsolete. Images "img1" to "img2" will
  be integrated. The starting phi value for the first image is given
  by the START subkeyword, and the oscillation angle per image is
  given by the ANGLE or OSC subkeyword. Note that for many detectors
  the phi values are read from the header information (Mar, MarCCD,
  ADSC, Raxis IV (NOT Raxis II), DIP) and so the START and ANGLE
  keywords are optional, but, if given, will override the values in
  the image header.


    If profiles are being accumulated (this is the default, but see
  PROFILE ACCUMULATE for more details) then the images will be
  processed in blocks of size "nblock" images. By default the program
  will calculate a value (between 6 and 13) so that there are not just
  one or two images left in the final "block", so under most
  circumstances the user should not need to specify "nblock" explicitly.

        The "batch" numbers assigned to each image in the output MTZ
  file are, by default, the same as the image numbers (img1 TO img2).
  However, if the ADD subkeyword is used, the value "iseradd" is added
  to all the image numbers. This is important, as all batches MUST
  have a unique batch number for scaling and merging.

  More than one PROCESS keyword can be given, for example if the
  oscillation angle has changed during a continous phi rotation.
P
  eg PROCESS 1 TO 50 START 0 ANGLE 1.0
     PROCESS 51 TO 130 START 50 ANGLE 0.5
p
  If more than one PROCESS keyword is given WITHOUT an intervening RUN
  keyword, then providing the images are continuous in phi they will be
  integrated as a single RUN. If however, they are NOT continous in
  phi, they will be treated as separate RUNS (so that profiles will
  be accumulated separately for images in the two runs, and there will
  be no postrefinement for the last image of the first run). All
  reflections will, however, be written to the same MTZ file, but the
  generate file will be re-initialised for the second run.

  If a RUN keyword separates the PROCESS keywords, then the images
  will be integrated independently, but again, unless a new HKLOUT
  keyword is given, all reflections will be written to the same MTZ
  file.

  If no START keyword is given, the phi values from the image header
  will be used. If there are no phi values in the image header an
  error message will be given. If a START keyword but no ANGLE keyword
  is given, the oscillation angle will be taken from the header, but
  the phi value specified by the START keyword will override that in
  the header. 

  If the ANGLE keyword is given, then a START keyword MUST also be
  given.

2 ADD
  ADD <iseradd>

  Value added to each image number to give the BATCH number in the 
  output MTZ file

2 ANGLE
  ANGLE <oscrange>

  Angular range for each image in a BLOCK (synonymous with OSC)

2 BLOCK
  BLOCK <nblock>

  Number of images in a block for post-refinement or integration.

2 OSC
  OSC <oscrange>

  Angular range for each image in a BLOCK (synonymous with ANGLE)

2 START
  START <phist>

  Starting angle for a block of images


1 PROFILE

  This keyword is used to set parameters associated with profile
  fitting, including definition of the areas over which the standard
  profiles will be determined, optimisation of the measurement box
  (background rim parameters and overall box size), parameters
  determining which reflections will be included in the standard
  profiles and whether or not weighting is used both in the formation
  of the profiles and in fitting individual spots.

  For more information type More.

2 More

  1) Definition of ares for standard profiles  

  Normally a number of different standard profiles will be determined
  for different areas on the detector. These areas are by definition 
  rectangular, and are defined by a set of lines parallel to the
  detector X and Y directions. 

  By default for high resolution data (greater than 2.5A) the program
  will generate 6 lines in each direction, giving 5*5=25 standard
  profile areas of which 4 will be outside the circular resolution
  limit, leaving 21 active profiles. For resolutions below 2.5A it
  will generate 4 lines giving 3*3=9 standard profiles.  However the
  user may define his own areas on the detector by supplying the
  coordinates (in mm) of the defining lines (XLINES,YLINES).

  The overall measurement box size for each of the standard profiles
  will be determined by expanding the peak region of the measurement
  box determined for the central region of the detector to allow for
  the increase in spot size due to obliquity (this depends on the
  Bragg angle and the nominal detector thickness). However the optimal
  background parameters (NRX,NRY,NC) will be determined separately for
  each of the standard profiles.

  2) Optimisation of the measurement box parameters

  By default the program will now automatically determine the best
  measurement box parameters. Using the supplied overall dimensions
  NX,NY it will first optimise the rim and corner parameters NRX,NRY
  and NC. It will then check that there at least twice (RATIO) as many
  background pixels as peak pixels, and if not it will increase NX and
  NY until this condition is satisfied (unless the proximity of
  neighbouring spots prevents this).

  The algorithm employed is as follows. The parameters NRX, NRY, NC
  are varied in turn and the value giving the highest ratio of
  Intensity to sigma(I) is found. This is the notional optimum
  value. The total intensity for this value is checked, and if it is
  less than the maximum intensity (for any value) by more than a
  factor 0.01 (TOLERANCE), then the rim parameter (NRX, NRY or NC) is
  increased by up to IBOUND pixels to try and regain this lost
  intensity.  For the first round of this iteration, when optimising
  NRX and NRY, NC is set to zero. NRX and NRY are varied from 1 to a
  maximum value which would give a peak dimension of 5 pixels. NC is
  varied from the smaller of NRX and NRY up to the smaller of (NX-2)
  and (NY-2).

  Two rounds of optimisation are performed, the second round using the
  results of the first.

  The optimisation is first carried out on the average spot profile
  for the central region of the detector. The optimisation of the
  overall dimensions is only carried out at this stage. The background
  parameters are optimised for ALL the standard profiles. 

  The value of TOLERANCE actually used in the optimisation depends on
  the position of the profile being optimised. For the central profile
  (that containing the direct beam position) the value TOLMIN is
  used. This is increased in a linear fashion as the profiles get more
  remote from the centre, up to a maximum value TOLMAX for the outer
  boxes.

  The parameters are reoptimised for each new BLOCK of images.

  The parameters IBOUND and TOLMIN,TOLMAX can be set using subkeywords
  BOUNDARY and TOLERANCE on the PROFILE keyword: 
P
  PROFILE BOUNDARY 3
  TOLERANCE 0.01 0.01
p
  corresponds to the default values.

  This optimisation can be turned off using subkeyword NOOPTIMISE:
  PROFILE NOOPTIMISE

  This will still optimise the raster parameters using the the average
  spot profile for the centre of the image, and use these parameters
  to CALCULATE appropriate raster parameters for each of the standard
  profiles (the peak size is expanded based on obliquity of incidence)
  but it will not optimise the individual standard profiles used in
  integration.  
  To suppress optimisation of the raster parameters completely, use
  keywords:
P
  PROFILE NOOPTIMISE ATALL
p
  Optimisation of the overall dimensions (NX and NY) can be suppressed
  with subkeyword FIXBOX:
P
  PROFILE FIXBOX
p
  The desired ratio of background to peak pixels (BGPKRAT) can be set
  with subkeyword RATIO:
P
  PROFILE RATIO 3.0
p

  3) Formation of the standard profiles

  The standard profiles themselves are formed from weighted sums of
  contributing reflections, so as to minimise the weighted fit of all
  contributing reflections to the derived profile. The weghts are
  based on counting statistics. Alternatively if NOWSUM keyword is
  given the profiles are formed by simply adding together all
  contributing reflections (with no weighting). If partials are being
  included in the profiles (use subkeyword PARTIALS) then the NOWSUM
  option should be used.

  4) Evaluation of the profile fitted intensity

  The profile fitted intensity is derived by a weighted fit of the
  appropriate standard profile plus a background plane to the pixel
  values of the spot. This fitting is done over ALL pixels of the
  measurement box excluding only background pixels rejected as
  belonging to a neighbouring spot (done automatically) or those
  rejected as outliers (using the BGSIG parameter on BACKGROUND
  keyword).ie there is NO destinction between peak and background at
  this point, and the standard profiles cover the entire measurement
  box, not just the peak region. If the keyword NOWEIGHT is given,
  then the fitting of the profile to the spot will be unweighted and
  the fit will only use the peak area of the measurement box, while
  the background will only be determined from the background
  pixels.
  The actual profile used for any given reflection is normally
  a weighted sum of the standard profiles for the areas surrounding
  the spot, weighted according to distance from the spot. This means a
  separate profile is evaluated for every reflection. This can be
  suppressed using keyword NOVARIABLE.

2 NOOPTIMISE
  NOOPTIMISE (ATALL)

  Optimise raster parameters for the average spot profile (in centre
  of image) and use these to calculate appropriate raster parameters
  for each of the standard profiles, but DO NOT optimise the
  measurement box parameters for each of the standard profiles.(See
  More for details)

  If the ATALL SUBkeyword is included, then no optimisation at all
  will be done, and the input raster parameters (RASTER KEYWORD) will
  be used, although the peak area of the spot will still be expanded
  to allow for obliquity of incidence on the detector when setting the
  raster parameters for the standard profiles.

  Default is to optimise parameters

2 TOLERANCE

  TOLERANCE <TOLMAX>
  or
  TOLERANCE <TOLMIN TOLMAX>

  When optimising the measurement box parameters NRX,NRY,NC, if
  necessary decrease these parameters (thus making the peak area
  larger), by up to IBOUND pixels (set by BOUNDARY) from those giving
  the maximum I/sig(I), so that the integrated intensity of the
  profile is not less than a fraction TOL of the maximum intensity for
  any value of these parameters. The value of TOL is calculated from
  TOLMIN and TOLMAX so that it is TOLMIN for the centre of the
  detector and varies linearly up to TOLMAX for the outer regions.
  
  Default 0.01 0.01, unless SYNCHROTRON keyword is included inwhich case it
  is 0.01 0.03. For very close spots or strong diffuse scatter it may
  be necessary to increase TOLMAX to 0.02-0.04

2 BOUNDARY

  BOUNDARY <IBOUND>

  When optimising the measurement box parameters from the values
  giving the maximum I/sig(I) to try to recover lost intensity (see
  TOLERANCE) only allow the parameters NRX,NRY,NC to change by a
  maximum of IBOUND pixels.

  Default: 6

2 FIXBOX 

  Do not optimise the overall measurement box dimensions NX,NY.

  Default is to optimise to achieve the desired ratio of peak to
  background pixels (set by RATIO)

2 RATIO

  RATIO <BGPKRAT>

  When optimising the OVERALL dimensions of the measurement box
  (NX,NY) increase these parameters so that the ratio of background
  points to peak pixels is at least BGPKRAT.

  Default: 2

2 STOP

  STOP <FRACREJ>

  When optimising the OVERALL dimensions of the measurement box
  (NX,NY), stop expanding in a particular direction (X or Y) if a
  fraction greater than FRACREJ of the new background pixels are rejected.
 
  Default: 0.5

2 HIGHRES
  Set up the binning for the standard profiles as in the default case
  for data to a resolution higher than 2.5A, giving 21 active profiles
  across the detector, irrespective of the true resolution of the
  data.

  Note that LOWRES,HIGHRES and XLINES/YLINES are mutually exclusive.

  Default: Set up number of bins depending on the resolution of the
  data, 21 for resolution higher than 2.5, otherwise 9.

2 LOWRES
  Set up the binning for the standard profiles as in the default case
  for data to a resolution lower than 2.5A, giving 9 active profiles
  across the detector, irrespective of the true resolution of the
  data.

  Note that LOWRES,HIGHRES and XLINES/YLINES are mutually exclusive.

  Default: Set up number of bins depending on the resolution of the
  data, 21 for resolution higher than 2.5, otherwise 9.

2 XLINES

  XLINES <XLINE1,XLINE2,.....>

  This defines the boundary lines for setting up the areas over which
  the standard profiles will be determined. (See More). XLINE1,XLINE2
  etc are the positions of the lines parallel to the Y axis. A set of
  lines parallel to the X axis must also be defined using YLINES.
  These lines MUST include both edges of the detector.
  eg To get 16 equal areas on a small Mar scanner use:
  XLINES 0 45 90 135 180 YLINES 0 45 90 135 180

  To get 25 equal areas on a large (30cm) Mar scanner use:
  XLINES 0 60 120 180 240 300  YLINES 0 60 120 180 240 300

  Default: Set up 21 active profiles for resolutions higher than 2.5A,
  otherwise 9 active profiles.

2 YLINES

  YLINES <YLINE1,YLINE2,.....>

  This defines the boundary lines for setting up the areas over which
  the standard profiles will be determined. (See More). YLINE1,YLINE2
  etc are the positions of the lines parallel to the X axis. A set of
  lines parallel to the Y axis must also be defined using XLINES.
  These lines MUST include both edges of the detector.

  eg To get 16 equal areas on a small Mar scanner use:
  XLINES 0 45 90 135 180 YLINES 0 45 90 135 180

  To get 25 equal areas on a large (30cm) Mar scanner use:
  XLINES 0 60 120 180 240 300  YLINES 0 60 120 180 240 300

  To get a single profile over the whole of a small Mar detector
  XLINES 0 180 YLINES 0 180

  Default: Set up 21 active profiles for resolutions higher than 2.5A,
  otherwise 9 active profiles.


2 NOVARIABLE

  If present, the standard profiles will be determined in the usual
  way, but in the integration pass the standard profile for the area
  in which the spot lies will be used rather than a weighted sum of
  the four standard profiles for the areas surrounding the reflection.

  Default: Use a variable profile for every spot (See More).


2 NOWSUM

  Do not use weights in forming the standard profiles. Only use this
  if you are including partials in the formation of the profiles.. See
  More

  Default: Use weights

2 NOWEIGHT

  Do not use weights in fitting the standard profile to an individual
  reflection. NOT RECOMMENDED. See More

  Default: Use weights 

2 NOINTERPOLATE

  Do not interpolate pixel values onto the calculated reflection
  positions. NOT RECOMMENDED

  Default: DO interpolate.


2 ACCUMULATE

  Build up the profiles using more than one image.  For full
  description see More. To turn off use NOACCUMULATE 

  DEFAULT:
  Accumulate profiles over all images given in PACK keyword or in the
  BLOCK subkeyword on the PACK card (when using POSTREF option)

3 More
        The  ACCUMULATE  option  allows  evaluation  of  the  standard
  profiles  using  reflections  from  more  than one image.  This has the
  advantage that the number of reflections contributing to each standard
  profile  is increased, which should improve the quality of the profile
  particularly for the high resolution profiles near the  edges  of  the
  image.  The use of this option is valid ONLY if two conditions are met:

       1) There is no change in the spot shape over the  images  included
  (radiation damage can lead to significant changes in spot shape).

       2) There is no appreciable  crystal  slippage  which  could  lead
  either  to  positional  errors  in  the  predicted  spot  positions or
  misclassification of fully versus partially recorded reflections.

          The flow of the program is rather different to  that  normally
  followed, and is outline below:

       1) FIRST PASS

       a) Perform positional refinement   of  the  first
  image  specified  on  the PROCESS keyword and write the pixel values for
  all spots to the scratch file 'SPOTOD'.

       b) Repeat (a) for all other images  specified  on
  the  PROCESS keyword,  in  each  case  writing  the  spot pixel values
  sequentially   to   the   file   'SPOTOD'.     Four    parameters,
  XCEN,YCEN,NREF  and  MAXR are saved for each film as they are required
  in the final integration pass.

       2) SECOND PASS

       a) The 'SPOTOD' file is rewound  and  the  standard  profiles
  formed from ALL the images processed during the first pass.

       b) These profiles are used to evaluate intensities  and  standard
  deviations for all images.



2 NOACCUMULATE
   Do not accumulate profiles.

2 SAVE

  Write profiles to logical file PROFILE for processing other images.

  DEFAULT:  Do not save profiles.

2 READ

  Do not calculate standard profiles but read them from the logical
  file PROFILE.

  DEFAULT:  Do not read profiles.

2 NOPRINT

  Switch of printing the standard profiles.

     DEFAULT:  Do print profiles.

2 NOCHANGEMASK

  Do not update the background mask based on the calculated overlap of
  neighbouring spots. NOT RECOMMENDED

  DEFAULT: Do reject overlapped background pixels.

2 BGSIG 

   BGSIG <X>

   Background points with densities deviating by more than X * r.m.s.
  variation in the background plane are rejected and the background
  plane recalculated (see keyword BACKGROUND for a fuller description
  of how the best background plane is determined. These points will
  also be rejected from the background mask used to integrate
  individual refelctions (see CHANGEMASK keyword).  

  Because the program now automatically masks out pixels which
  actually belong to the peak region of neighbouring spots, this
  option is no longer required and BGSIG shpuld be left at its default
  value of 99.0. To change this value you need to include the
  subkeyword MODIFY: **** THIS IS NOT RECOMMENDED ****

  BGSIG MODIFY 25.0

  DEFAULT  99

2 ISDR 

   ISDR <N>

  Reflections with a peak values less than N * r.m.s.  variation in
  the background density are rejected when forming the standard
  profiles as being too weak.

  DEFAULT 2

2 CUTOFF 

    CUTOFF <x>

  Reflections with more than NOVPIX pixels with values greater than
  'x' are rejected as overloads.  If instrumental errors are believed
  to be significant at high pixel values, this parameter can be used to exclude
  affected reflections from the profiles. (NOVPIX is set on the
  OVERLOADS keyword using subkeyword NOVER and defaults to 0)

  DEFAULT: Image plate 65000  (If keyword SITE LMB or SITE
  EMBL SCRn given, default cutoff is 31999). If the CUTOFF for
  integration is changed (OVERLOAD CUTOFF keywords), but the cutoff
  for profile formation is not, then the cutoff for profile formation
  is set to the smaller of the default value and that given for spot
  integration.
                               
2 NREF

   NREF <N>

   A standard profile will be rejected (and therefore averaged) if less
   then 'N' reflections contribute to the profile. NREF should NEVER
   be set to less than 5, as this will give inaccurate estimates of
   the standard deviation of the profile fitted intensity (which is
   derived from the fit of the standard profile to the reflection).

  DEFAULT 10

2 RMSBG

    RMSBG <X>

  A standard profile will also be rejected if the r.m.s.  variation in
  the background after scaling the profile in the range 0-255 is
  greater than 'X'. (Note that the profiles actually used for
  integration are scaled to a maximum value of 10,000 NOT 255).

  DEFAULT 10.0

2 PARTIALS

  If present, partially recorded reflections will be included in
  forming the standard profiles. 

  It is normally only used when the combined mosaic spread and beam
  divergence is comparable to the oscillation angle, so there are very
  few fully recorded refelctions, and the ADDPART option cannot be
  used because of variations in incident beam intensity or instability
  in the detector (eg variations in ROFF, TOFF or CCOMEGA).  If
  partials are included, then the weighting for forming the standard
  profiles is automatically turned off, as it is no longer
  appropriate.

  DEFAULT: Exclude partials from profiles

2 OVERLOADS

  If present, the intensities of overloaded reflections (see keyword
  OVERLOAD for a definition of what is considered an overload) will be
  estimated by profile fitting that part of the peak which is not
  overloaded.  These intensities will clearly have a larger standard
  deviation than non-overloaded reflections, and will therefore give
  rise to a higher R-factor than non-overloaded reflections of similar
  intensity. At present, the "integrated" intensity is set equal to
  the profile fitted intensity for these reflections.  This will lead
  to an artificially low value for the R-factor between profile fitted
  and integrated intensities. These reflections are not flagged in any
  way in the generate file.

  Default: Reject overloaded reflections

2 EDGE

  If present, reflections whose measurement box lies partly outside
  the scanned area will be integrated by profile fitting, providing at
  least half of the pixels lie within the scanned area. This feature
  is still being evaluated for its usefullness.  If this option is to
  be used, it is ESSENTIAL to dump some of the edge reflections
  (keywords: DUMP EDGE) to check that the pixel values at the very
  edge of the scan are sensible. Some detectors seem to give a "rim" of
  anomalously high values.

  Default: Reflections with any pixels outside the scanned area are
  rejected.


2 WDLIM1
  WDLIM1 <PKWDLIM1>
  Controls rejection of individual pixels from the profile
  fitting. Pixels are rejected if SQRT(W*DEL**2) is greater than
  PKWDLIM1 where DEL is the difference between the scaled standard
  profile and the background corrected pixel value and W is the
  expected variance of the pixel value. However this test is only
  applied on the second pass of pixel rejection...see WDLIM2 below.
  Default 20.0

2 WDLIM2
  WDLIM2 <PKWDLIM2>

  Similar to WDLIM1, but this cutoff is applied to pixels that are
  adjacent to a pixel overlapped by a neighbouring spot. It is
  intended to catch pixels which are artificially high due too a very
  strong neighbouring spot "spilling over" into the current spot. Only
  pixels in this class are rejected on the first pass of pixel
  rejection, then on the second and subsequent passes those failing
  the WDLIM1 test are also rejected (up to 33% of the peak pixels
  only).
  Default 20.0

2 DISCRIMINATE
  DISCRIMINATE <DISCRIM>

  This option can be used to eliminate reflections with strong
  neighbours from formation of the standard profiles, thus minimising
  corruption of the standard profiles due to counts from a strong
  neighbour "spilling" over into the current spot. Spots are rejected
  from the profiles if the background corrected pixel count in the
  central pixel of the spot is less than DISCRIM*MAXBGP where MAXBGP
  is the maximum bbackground corrected pixel counts for any pixel
  within the measurement box (but outside the peak of the current
  spot). Thus a value of 2 for DISCRIM would mean that only spots
  which were at least twice as strong as their immediate neighbours
  will be included in forming the profiles.
  Default 0.0


1 RASTER
        
  RASTER < NXS NYS NC NRX NRY >
  

      These five parameters define the spot measurement box.  The X
    direction is the "slow" direction in the scanned image, the Y
    direction is the "fast" direction. Thus for the Mar scanner, X is
    parallel to the rotation axis, Y is vertical.  NXS, NYS (odd
    integers) define the overall dimensions of the rectangular array
    of pixels for each spot. NRX, NRY are the number of columns or
    rows of points in the background rim and NC the corner background
    cut-off which corresponds to a half-square of side NC points.
P
    eg    RASTER 15 15 7 2 3
p
    Default: If no RASTER keyword is given, the program will
    automatically set up appropriate values. It will first determine
    the median size of spots in the centre of the image (in fact,
    around the direct beam position). The parameters defining where
    this search is done can be set using the SPOTS keyword. NXS and
    NYS are then set to twice the spot size plus 5 pixels, NRX and NRY
    are set to half the spot size plus 2 pixels, and NCS is set to the
    mean of the spot size in X and Y plus 4.

    Note that the parameters NC,NRX,NRY will automatically be
    optimised after detector parameter refinement for the first image
    in every BLOCK of images (unless PROFILE NOOPT ATALL has been
    set). However, the overall box size will NOT be changed, unless
    the keywords PROFILE NOFIXBOX are supplied, in which case the
    overall box size will be optimised also as described under the
    optimisation of ther measurement box parameters section (under
    PROFILE).

1 REFINEMENT

  The transformation between predicted reflection positions and their
  observed positions (centroids) in the digitised image is refined for
  each image separately. The refined parameters are the camera
  constants, and "distortion" parameters: crystal to detector distance
  (XTOFRA), detector Y axis scale factor (YSCAL), tilt and twist of the
  detector. In addition for detectors that perform a spiral scan (Mar
  Research and Mac Science) a radial offset (ROFF) and a tangential
  offset (TOFF) are refined (these allow for mechanical misalignment
  in the detector). There is the option to refine radially dependent
  offsets RDROFF and RDTOFF, but by default these parameters are NOT
  refined.. All these parameters may be assigned initial values with
  the CAMCON and DISTORTION keywords. The refinement is initially
  performed using strong reflections from the centre of the image, and
  then goes on to use reflections over the entire image.

  The contribution of each selected spot to the positional refinement
  is weighted by the estimated error in its centroid position (based
  on counting statistics). Thus strong spots will get a higher weight
  than weak ones. The refinement (by default) minimises the weighted
  residual. Both the weighted and unweighted residuals are printed in
  the output. The weighted residual should always be close to unity
  irrespective of the strength of the image or size of the spots. The
  unweighted residual will be greater for weak images or large spot
  sizes.

      The selection of reflections, which parameters are refined and
  some aspects of the refinement are controlled by the following
  keywords:


2 FIX

  This allows individual parameters to be kept fixed during the
  positional refinement. Any of the following parameters may be fixed:
  XCEN, YCEN, CCOMEGA, YSCALE, XTOFRA, TILT, TWIST, ROFF, TOFF, RDTOFF,
  RDROFF.
P
  eg REFINEMENT FIX ROFF TOFF
p
  will prevent refinement of parameters ROFF, TOFF (radial and
  tangential offsets).
 
  For very weak images, it may be best to fix ROFF and TOFF for images
  collected on Mar or Mac Science detectors.

  Default: For Mar and Mac Science scanners all parameters listed
  above are refined with the exception of RDROFF and RDTOFF, which
  must be explicitly turned on with a REFINEMENT FREE keyword. The
  radially dependent distortion parameters are only refined when using
  data from the entire image.
  For all other detectors, ROFF,TOFF, RDROFF, RDTOFF are not
  appropriate and are not refined.

2 FREE

  This allows parameters that are normally fixed to be refined. The
  only parameters in this class are the radially dependent radial and
  tangential offsets (RDROFF and RDTOFF). These parameters should only
  be refined when working with strong high resolution images. They
  will not normally make any detectable difference to the processing,
  unless there is a problem with the alignment of the detector optics.
P
  eg REFINEMENT FREE RDTOFF
p


2 NREF
  NREF <MINREF>
  Sets the minimum number of reflections for refinement of the
    detector parameters using data from the central region of the
    detector. If fewer than this number are found, the program will
    abort. When processing data from small moleclues it may be
    necessary to reduce MINREF from its default value of 20.
    Default: 20

2 NOWEIGHT

  Minimise the unweighted residual rather than the weighted one.

  Default: Minimise the weighted residual


2 REJECTION 
  
  REJECTION <REFREJ>
  Controls rejection of individual reflections from refinement. If the
  error in calculated position is more than REFREJ*(estimated error in
  centroid position) the reflection is rejected.

  Default: 3.0

2 NSIG

     NSIG <ISIG>

  When refining using reflections from the central region of the image
  (CENTRS), those reflections with an observed standard deviation in
  the pixel values less than ISIG times a calculated standard
  deviation based on counting statistics are rejected as too weak (see
  description of CENTRS).  This type of comparison is used rather than
  a test on integrated intensity because the measuring box is not used
  when refining the central region (unless USEBOX is specified), so as
  to give a larger radius of convergence.  If the spots are small
  compared to the size of the measurement box it may be necessary to
  change the default value.  

  DEFAULT: 6 but this is reduced automatically if too few spots found.

2 INCLUDE

  INCLUDE [PARTIALS <PTMIN> OVERLOADS]

  This keyword controls selection of reflections for positional
  refinement, both in the central region of the image (CENTRS) and the
  outer region (SEEKRS or NEXT). Normally only fully recorded,
  non-overloaded reflections are allowed, unless there are relatively few
  fully recorded reflections (see the FULLFRAC subkeyword) in which case
  partials are automatically included. 

  The automatic inclusion of partials can be overridden by using the INCLUDE
  subkeyword (and the FULLFRAC subkeyword) with the appropriate subkeyword. 

  If using ADDPART then only partials at the end of the oscillation
  range will be selected and the other "half" of the partial will
  automatically be added in from the next image, so the centroids of
  these reflections will be determined as accurately as those of fully
  recorded reflections (provided the detector origin is indeed fixed).

  If this is not the case, partials from both the beginning and end of
  the oscillation will be selected but only if their degree of
  partiality is greater than <PTMIN> (defaults to 0.8 but can be changed
  by giving an appropriate value after the subkeyword PARTIALS:
P
  eg REFINEMENT INCLUDE PARTIALS 0.5 
p
  will include any partial more than 50% recorded)

  The program will now automatically reduce PTMIN


  Overloads are ALWAYS excluded from the average spot profile.

  ***** IMPORTANT ****

  If partials are used for positional refinement, and ADDPART (qv) is
  not being used, then by default partials will ALSO be included when
  forming the standard profiles for integration. If you do not want
  this to happen, include the keywords: PROFILE FULLS

  Default: Partials are not used unless the fraction of fully recorded
  reflections is less than that specified by the FULLFRAC subkeyword, or the
  program cannot find enough fully recorded reflections for positional
  refinement, in which case partials are automatically included.

2 FULLFRAC

  FULLFRAC <frac>

  This governs whether partially recorded reflections are included in
  refinement of the detector parameters (see also INCLUDE subkeyword for
  more details). For the FIRST image to be processed, if the fraction of
  reflections generated on that image that are fully recorded is less than
  "frac", then partials are automatically included. 

  Unless the PROFILE FULL keywords are given, this will also mean that
  partials will be included in formation of the standard profiles. Note that
  ONLY the first image is used when setting the decision on whether or not
  to include partials.

  Default value for "frac": 0.333

2 LIMIT 

     LIMIT <x>

  When refining the detector and distortion parameters using
  reflections from the central region of the image (CENTRS), only
  reflections with X and Y  coordinates less than "x" mm are
  used.  

  DEFAULT 45mm 


2 RESIDUAL
     RESIDUAL <x>

  Sets the maximum allowed final residual in  CENTRS  (ie  for
  central region of the image) or in SEEKRS (ie refinement over
  the whole image) if processing is to  continue.  

  DEFAULT 6.0

2 CYCLES 
     CYCLES <NCYCLES>
  Sets the number of refinement cyles in CENTRS.  For  problem
  images,  eg when spots are very close along the rotation axis
  it can be beneficial to increase NCYCLES.

  DEFAULT 2


2 IMIN

   IMIN <SDFMIN [SDFINC]>

  Controls selection of refinement spots in the outer regions of the
  image (SEEKRS). Reflections with an integrated intensity greater
  than SDF_BIN*(SIGMA(I)) will be selected, where SIGMA(I) is
  calculated from counting statistics.  SDF_BIN is set to SDFMIN in
  the outermost regions of the image and incremented by SDFINC for
  each bin working towards the centre.  If only SDFMIN is given,
  SDFINC defaults to SDFMIN/2 

  Default: 40, 10, but these values are automatically reduced by
  program if too few spots found.

2 BGREJ

   BGREJ <BGFREJ>

  Spots in which more than a fraction BGFREJ of the background pixels
  have been rejected as outliers from the background plane will be
  rejected from refinement.

  Default: 0.5
   
2 GRADIENT

  GRADIENT <GRADMAXR >

  Spots with a ratio of the either background plane gradient (a or b)
  to the average background (ie a/c or b/c) greater than GRADMAXR wil be
  rejected from the refinement.

  Default: 0.03

2 NOUSEBOX

   In the initial stages of refinement in the central region of the
  image, the mask defining the peak and background regions of a spot
  will not be used, but the centroid of the whole measurement box is
  found. This improves the convergence of the refinement, but will not
  work if spots are close together.

  Default: DO use the peak/background mask even in the first stages of
  refinement in the central region of the detector.



1 REJECTION

   This sets criteria for flagging reflections as "Badspots" which are
  rejected from further processing stages (by flagging with a standard
  deviation of -9999). Badspots are not included in the final
  statistics. All badspots are listed in the ouput, and if running
  online there is the option of inspecting each badspot and deciding
  whether or not it should be rejected.  The pixel values for badspots
  can be printed in a batch job using the PLOT subkeyword (see below).

2 PKRATIO

  PKRATIO <X> [ACCEPT]

  Reflections with a PKRATIO greater than X will be flagged as
  "badspots" with the message "Poor profile fit".

  See Documentation, Errors for definition of PKRATIO.

  Treatment of reflections which fail ONLY the PKRATIO test (and not
  BGRATIO, GRADMAX etc) depends on the presence or absence of the
  subkeyword ACCEPT. If not present, then the reflection will be
  rejected completely. If ACCEPT is present, then for these
  reflections the profile fitted intensity and standard deviation will
  be set to the summation integration intensity and standard
  deviation. 

  The choice between these two options depends on how the data is to
  be used. For molecular replacement, completeness of the data is
  crucial, and since reflections failing the PKRATIO test are often
  the strongest ones, it is better to take the summation integration
  estimate even if this is slightly in error. However, for multiple
  isomorphous replacement work, it may well be best to reject all
  "suspect" reflections entirely (ie do NOT use the ACCEPT keyword).

  In fact, if a significant number of strong reflections are being
  rejected by the profile fit, it may be best to increase the value of
  PKRATIO slightly (eg up to 4).

  Default 3.5

2 BGRATIO

   BGRATIO <X>

   Reflections with a BGRATIO greater than X will be flagged as
  "badspots". See Documentation, Errors for definition of BGRATIO.
  Default 3.0

2 GRADIENT

   GRADIENT <GRADMAX >

  Spots with a ratio of the either background plane gradient (a or b)
  to the average background (ie a/c or b/c) greater than GRADMAXR wil be
  rejected from the refinement.

  Default: 0.03

2 MINBG

   MINBG <NBGMIN> 

   Following rejection of outliers from the background plane,
  reflections with less than NBGMIN background pixels will be
  rejected.

   Default 10

2 PLOT

   PLOT

    A number field of the pixel values will be printed for all
  rejected reflections but ONLY if the detector error factor has
  been assigned a value (DETECTOR keyword).


1 RESOLUTION
 
  RESOLUTION [ <lowres> ] < highres> > [CUTOFF <sigcut>] 
      [EXCLUDE <dmin dmax> ] [ANISOTROPIC <res1 res2 res3> [LOW <lowres>]]
      [EXCLUDE ICE] [EXCLUDE NONE]

        The resolution limits in angstrom. If this is not specified,
  the resolution is determined by the size of the detector (LIMITS
  keyword). The low resolution limit is optional, and if not given is
  determined by RMIN (the minimum radius). If both resolution limits
  are given, they can be given in either order.

        The CUTOFF subkeyword allows the resolution limit of
  reflections written to the MTZ file to be different for each
  image. The resolution limit is set as that resolution at which
  I/sigma(I) drops to below "sigcut". The I/sigma(I) for fully
  recorded reflections (if any) is used, otherwise partials.
  Default sigcut 0.0

        The EXCLUDE subkeyword allows specified resolution ranges to
  be excluded from measurement (eg to miss an ice powder ring). Up to
  10 different ranges can be given each with its own exclude
  keyword. eg. EXCLUDE 3.66 3.72 EXCLUDE 10.0 10.2

	EXCLUDE ICE specifically excludes six resolution range rings 
   which coincide with expected ice rings, i.e. 3.94 - 3.86, 
   3.71 - 3.63, 3.48 - 3.40, 2.71 - 2.63, 2.28 - 2.22 and 2.08 - 2.02A.

	The EXCLUDE NONE keyword unsets all excluded rings, and has been 
   added for use in automated procedures and the new Mosflm GUI.

   The ANISOTROPIC subkeyword allows a different high resolution limit
   to be specified for each reciprocal lattice axis. It overrides a
   previously given resolution limit, and is overidden by a subsequent
   resolution limit. 

   eg ANISO 2.5 3.2 5.0 will exclude reflections from processing which
   have a resolution higher than 2.5, 3.2 and 5.0 A along a*, b* and
   c* respectively.

   The LOW subkeyword can be used in combination with the ANISOTROPIC
   keyword to specify the low (isotropic) resolution limit; IT IS
   IGNORED unless it has been preceded by the ANISOTROPIC keyword.
P
   eg RESOLUTION ANISO 3.5 2.7 2.4 LOW 15.0
      RESOLUTION 22.5 1.7 EXCLUDE 3.6 3.8
p
1 RUN
   RUN (or GO, these are equivalent) This starts the program running,
  so should be given at the end of the keyword input (See "Example").
  If processing two discontinous segments of data, the first segment
  should be defined using the PROCESS keyword and all other necessary
  keywords, and this should be followed by a RUN keyword, followed in
  turn by a PROCESS keyword for the second segment, then another RUN
  keyword. 

  RUN (or GO) keywords are also used when doing a multi-part STRATEGY
  run.


1 SCANNER

  OBSOLETE: see the DETECTOR keyword.

1 SDMON
   SDMONITOR X
     When analysing agreement between symmetry related fully recorded
  (or summed partial) reflections, print warning message for thos
  reflections where any observation siffers from the weighted mean by
  more than "X" standard deviations.

  Default 5.0


1 SERIAL

   ** This keyword is OBSOLETE. Please use the PROCESS keyword. **

1 SEPARATION
  SEPARATION [ <dx dy> ] [CLOSE TRIM x OVERLAP x]

        The minimum separation (in mm) between spot centers in the detector X
    (slow) and detector Y (fast) directions.  Spots closer than these limits
    will not be integrated (they are displayed with red boxes in the display
    and described as spatial overlaps).
    
    Suitable values for dx and dy can be determined from the size of a strong
    reflection (use the graphics display to determine this).  Using too small
    a value will result in problems in processing !  When testing for
    overlaps, the spot is modelled as having an elliptical shape with full
    axis sizes of dx, dy mm in X and Y.

    Default: If no separation is given, the spot size is determined by the
    program and the minimum separation set initially to the spot size plus two
    pixels. However, after detector parameter refinement for the FIRST image
    of each "block" of images, using only spots in the centre of the image,
    the program determines the optimum measurement box parameters. The size of
    the spots as determined by the size of the "peak" region of the
    measurement box following this optimisation is used to update the estimate
    of the minimum separation, and in this case the separation is set to the
    actual spot size (not adding two pixels). If running interactively, the
    display will automatically be updated to reflect the number of overlaps
    with the modified separation. Note that this is ONLY done if no separation
    parameters are supplied on the SEPARATION keyword. Also note that since
    the optimum measurement box parameters are affected by the PROFILE
    TOLERANCE values (see under PROFILE) then the calculated separation
    parameters will also be affected.

    In most cases the automatic determination of the spot separation works
    very well and there is no need to supply a SEPARATION keyword. However,
    when the spots are very close, it may be an advantage to override the
    values calcualted by the program.

    CLOSE, TRIM and OVERLAP modify the way in which overlapping spots
    are treated. CLOSE is the recommended option for very dense
    patterns. 

    USE OF "TRIM" AND "OVERLAP" ARE NO LONGER RECOMMENDED.

    Note that if no separation (dx,dy) values are given, the program
    will automatically determine if the "CLOSE" option should be
    used. If the closest possible spot separation on the detector
    (determined from the lattice type, cell dimensions and crystal to
    detector distance) for any crystal orientation is less than the
    spot size plus a safety margin of 2 pixels (this can be changed by
    keywords SPOTS SAFE) then the "CLOSE" option is invoked.
   
    **** IMPORTANT ***
    If the "CLOSE" option is being used, then it is important that all
    images in the dataset are also integrated using the "CLOSE"
    option. Use keywords SEPARATION CLOSE to ensure this happens.

2 More

  a) Standard treatment (without CLOSE option)
  ============================================  
     The treatment of overlapping spots is worked out independently
  for each area on the detector for which an standard profile is
  evaluated (by default there are 9 such areas for data below 2.7A
  resolution, 22 active areas for higher resolution).

  Each area has a fixed size measurement box and a fixed
  peak/background mask. For every background pixel in this mask, if
  there is a single spot (in this area) for which this pixel is
  overlapped by the PEAK of a neighbouring spot, then that pixel is
  rejected from all calculations for ALL reflections in this area
  (this is indicated by a * when the standard profiles are printed).
  The size of the peak region is determined automatically by the
  profile optimisation (See under PROFILE or the User Guide).

  During integration,a "best" profile is evaluated for each reflection
  as a weighted mean of the standard profiles for the surrounding
  areas, and all background pixels which are rejected for ANY of the
  contributing profiles are rejected.

  These criteria are very stringent, and may lead to the situation
  where almost all background pixels are rejected for some standard
  profiles, because too large a rotation angle per image has been
  used. In order to cope with such cases, the criteria can be relaxed
  using the TRIM and OVERLAP.

  **** IT MUST BE REALISED THAT THIS WILL LEAD TO SOME DETERIORATION
  IN DATA QUALITY. IT IS FAR BETTER TO USE A SMALLER ROTATION ANGLE OR
  BETTER COLLIMATION TO REDUCE THE NUMBER OF OVERLAPS IF THIS IS
  POSSIBLE ****

  b) Special treatment when CLOSE keyword is given
  ================================================ 

  If the CLOSE subkeyword is supplied, then the overlapped background
  pixels are determined for every individual spot, using the known
  coordinates and sizes of the neighbouring spots. This will, in
  general, provide more background pixels than the standard
  treatment. If there are too few background pixels (determined by
  keywords PROFILE RECOVER and REJECTION MINBG) then the peak area of
  neighbouring spots will be "trimmed" back until sufficient
  background pixels are obtained. This "trimming" procedure is applied
  to the weakest of the neighbouring spots first, trimming back an
  increasing number of the neighbours until enough background pixels
  are found. This wil minimise the error introduced by using these
  pixels in determining the background.  


  Another consequence of using this procedure is that it is no longer
  appropriate to fit the sum of the scaled standard profile plus a
  background plane to all pixels (except overlapped ones) in the
  measurement box, because the profiles will contain contributions
  from neighbouring spots at pixels which may not be overlapped when
  considering any individual spot. Thus, the profile is only fitted to
  the peak region of the spot, and the background is fitted to all
  non-overlapped background pixels. (This is equivalent to using the
  PROFILE PKONLY keywords.)

  When using the CLOSE option, ADDPART CANNOT be used (because it is
  much more difficlut to implement). If CLOSE is specified, ADDPART is
  automatically turned off.

2 CLOSE

  Special treatment for very dense diffraction images. See More above.

2 TRIM
  TRIM <N>
      
     When checking for background pixels which are overlapped by a
  neighbouring spot, the effective size of the neighbouring spot will
  be decreased by 2*N pixels in X, Y and in the diagonal directions
  (by increasing the background rims in X and Y and the corner cutoff
  by N pixels). N is normally zero, giving N a positive value will
  reduce the number of background pixels rejected.

2 OVERLAP
  OVERLAP <N>

     Normally a background pixel is rejected if there is a single
  reflection for which that pixel is overlapped. However if this
  keyword is given this will only happen if there are at least N
  reflections for which that pixel is overlapped, thus reducing the
  number of rejected background pixels.

1 SHIFT 

   SHIFT <IX [IY]>

   This allows the measurement box to be placed so that the centre of
  the box is displaced from the calculated position of the spot (the
  calculated position will correspond to the centre of gravity).  The
  actual displacement is IX and IY pixels along detector X and detector
  Y respectively.  This option can be useful if the centre of gravity
  of the spot is not in the middle of the spot, as can happen with,
  for example, triangular or wedge-shaped crystals which are smaller
  than the incident beam.

1 SITE

   SITE sitename

   This keyword is used to set up the parameters for non-commercial or
   non-standard detectors installed at synchrotron sites.

   Sites currently allowed are: 
P
                                CHESS (for CCD or Fuji detectors)
                                EMBL  (for prototype image plate scanners)
                                SSRL   (for ADSC 2x2 CCD detector)
                                ALS    (for ADSC 2x2 CCD detector)
p
   This keyword will set up the physical limits of the detector (LIMITS), the
   default GAIN, the orientation of the detector and direction of the rotation
   axis (DETECTOR), the size of the image (SIZE), the pixel size (PIXEL), how
   values are stored in the image (DETECTOR TYPE) and the saturation values
   (OVERLOAD). All of these may be overridden by subsequent keywords if
   required.

   FOR ONLINE SCANNERS (Mar Reseach, R-axis, Mac Science) THIS KEYWORD IS NOT
   NORMALLY NECESSARY, UNLESS THERE IS SOMETHING UNUSUAL ABOUT THE DETECTOR AT
   A PARTICULAR SITE...for example the use of prototype scanners at EMBL.

   See the "DETECTOR" keyword for details of how to deal with non-standard
   detectors.


2 CHESS

   CHESS [A1 F1 F2] [FUJI [CCD [1K 2K 2KBINNED ADSC QUANTUM4]]]
  
   For SITE CHESS, the beamline (A1, F1 or F2) must be given, as the sense of
   the rotation axis is different on different stations (and also may change
   as stations are upgraded !!).

   This should be followed by the detector type, which can be FUJI
   (for the offline Fuji scanner) or CCD followed by the type of CCD.
   At present, 5 types of CCD are allowed. The first three are for the
   Gruner detector, which can operate in 1K, 2K or 2KBINNED modes, the
   fourth is the ADSC single module detector (ADSC) and the fifth is
   the ADSC 2x2 detector (QUANTUM4).
P
   eg SITE CHESS F1 CCD 2KBINNED
   or SITE CHESS A1 CCD ADSC
   or SITE CHESS F1 CCD QUANTUM4
   or SITE CHESS F2 CCD 1K
p
   Thre is NO default.

2 EMBL [SCR1 SCR2 SCR3]

2 SSRL 
  SSRL ADSC

   At present, only one option is available for SSRL, which is the
   ADSC 2x2 tiled CCD detector.
P
   eg SITE SSRL ADSC
p
2 ALS 
  ALS ADSC

   At present, only one option is available for ALS, which is the
   ADSC 2x2 tiled CCD detector.
P
   eg SITE ALS ADSC
p
1 SIZE

   SIZE <NREC  IYLEN> [ HEADER <NHEAD> [ BYTE <NBYTE>] ]

  The size of the image in pixels. The image should have NREC
  records each of IYLEN pixels ie there are IYLEN pixels in the fast
  changing direction in the image (defined as the detector Y axis), and
  NREC pixels in the slow changing direction (defined as detector X).

  For Mar images, this information is read from the image file header
  (if present) or the SITE specification for prototype detectors.
  
  For non-standard detector images, the number of header records can be
  given with the HEADER subkeyword. If the BYTE subkeyword is omitted,
  then these header records are assumed to be the same length as those
  in the image (ie IYLEN 2-byte words). However, the length of the
  header records (in bytes NOT words) can be specified with the BYTES
  subkeyword. The header records will be skipped (ie no information
  will be read from them).

   Default: Image size is set up automatically according to the
   information given on the DETECTOR keyword.


1 SPOTS

  SPOTS subkeywords 
  subkeywords: RMIN RMAX NSEARCH SCALE SAFE

  When finding spots for autoindexing, unless a THRESHOLD value has
  been given explicitly (see FINDSPOTS keyword), the program will
  determine a suitable value automatically. Spots will be located in
  the central region of the image between radii of RMIN and RMAX using
  the default value for the threshold. Providing at least NSEARCH
  spots are found, the mean intensity per pixel (xmean) and the mean
  spot threshhold (thrmean) of the NSEARCH strongest spots
  found is calculated. The average standard deviation in the
  background counts (sigavg) is also determined. The new threshold is
  then set to:

  THRESHNEW = (SCALE*xmean*(thrmean-THRESH*sigavg)*0.001)/sigavg + 3.0

  The same parameters (RMIN,RMAX,NSEARCH) are used when determing the
  median spot size used in setting up default values for SEPARATION
  and RASTER if these are not specified by the user.

2 RMIN
  RMIN <radmin>

    The minimum radius (in mm) used in looking for spots in centre of
  image. The search is centred on the direct beam position.

  Default: the greater of RMIN (as given by LIMITS keyword) or
  0.1*(radius of scanned area).

2 RMAX
  RMIN <radmax>

    The maximum radius (in mm) used in looking for spots in centre of
  image. The search is centred on the direct beam position.

  Default: radmin + 0.3*(radius of scanned area).

2 NSEARCH
  NSEARCH <nspots>

     The number of spots used in determining threshold.

  Default: 10

2 SCALE
  SCALE <scal>

     The scale factor used in determining the threshold.

  Default: 0.02

2 SAFE
  SAFE <npix>

     The test for the use of the "CLOSE" option during integration if
  this has not been specified explicitly makes use of a safety factor
  "npix". See the SEPARATION keyword for more details.

  Default: 2

1 STRATEGY

  STRATEGY  subkeywords
            subkeywords: AUTO ROTATE SEGMENTS SIZES START END PARTS
                         SPEEDUP ANOMALOUS NOTANOM ALTERNATE

   The STRATEGY option allows the design of a data collection strategy
  in a semi-automatic way for a single axis rotation camera. It
  requires all the parameters normally used to process a set of images
  (crystal symmetry, orientation, crystal to detector distance,
  wavelength, detector type, direct beam position). 
  
  Note that the type of detector (other than a Mar) MUST be set using a
  DETECTOR keyword. If invoked from the X-windows display, then it is
  not necessary to specify the size of the image for Mar detectors as
  this is read from the image header.

  Use keyword EXIT or END to end the strategy option.
  
2 AUTO
              Determine the starting phi angle and the phi rotation
              required to give a complete dataset (if possible from a
              single crystal setting), and give statistics on
              completeness and multiplicity.
              Do NOT use START or END with the AUTO keyword.

              If no START keyword is given, then AUTO mode will be the
              default, so if you give a ROTATE keyword there is no
              need to give an AUTO keyword

2 ROTATE
  ROTATE  <phirot>
              Only for use with the AUTO option. Restrict the total
              rotation to "phirot" degrees, rather than the full
              rotation for this Laue group and crystal setting.
              If no AUTO keyword has been given, ROTATE must be given
              before a SEGMENT keyowrd.

2 SEGMENTS
  SEGMENTS <nseg >
              Only for use with the AUTO option. Allow "nseg"
              discontinous segments of data to give a total rotation
              of PHIROT degrees. Unless specified explicitly with the
              SIZES keyword (see below) the segments will have
              approximately equal widths in phi.

              Default: If in AUTO mode and no SEGMENT keyword is
              given, then one segment will be assumed.
              eg ROTATE 60 is equivalent to AUTO ROTATE 60 SEGMENT 1

2 SIZES
  SIZES <size1,size2,size3...>

              The sizes for the "nseg" segments. If SIZES are given,
              then the "phirot" value given on the ROTATE keyword is
              ignored, and the total rotation is the sum of the SIZES.
              Default: Use approximately equal sizes with total "phirot"

2 START/END
  START <phistart>
  END   <phiend>
              As an alternative to AUTO mode, specify the start and
              end phi values to be used in generating the reflection
              list. Up to 10 different sets of START and END can be
              given on successive STRATEGY keywords. eg
              STRATEGY START 0  END 30
              STRATEGY START 35 END 60
              STRATEGY START 70 END 90

2 PARTS
  PARTS <nparts>

              If data have already been collected from other crystals,
              or some data has already been collected from the current
              crystal, set "nparts" to the total number of crystals
              (including the current crystal whose phi range is to be
              determined). See example below. This need only be given
              on the first STRATEGY keyword.

2 SPEEDUP
  SPEEDUP <factor>

    The calculation can be speeded up by reducing the cell size. This
  can be particularly useful for large cells (eg greater than 100A)
  although inevitably there will be some loss of accuracy. However,
  providing there are a few thousand reflections generated the loss in
  accuracy is small.  Suitable values for "factor " are 5-20.

  Default: The program will automatically calculate a suitable SPEEDUP
  factor based on the unit cell size if no value is specified.


2 ANOMALOUS

  ANOMALOUS
  Including this keyword will optimise the completeness of anomalous
  pairs rather than unique data.
  To cancel optimisation of anomalous data, use keyword NOTANOM

2 NOTANOM

  NOTANOM
  Use this to cancel optimisation of anomalous pairs rather than
  completeness of unique data.

2 ALTERNATE

  ALTERNATE
  When working with an orthorhombic spacegroup, the algorithm used to
  determine the best starting PHI value when using the AUTO mode is
  successful in approximately 90-95% of cases, but in some cases it
  fails. As a result, the best predicted completeness may be 3-4% less
  than it should be. To check this, a separate STRATEGY run should be
  performed including the keyword ALTERNATE. This chooses an
  alternative starting phi value and may, rarely, give a higher
  predicted completeness.

  This is only necessary for orthorhombic space groups.

2 More

    The rotation range (PHITOT) required to collect a complete dataset
  is determined from the crystal symmetry and orientation (eg 180
  degrees for Laue group P2/m if rotating about the b axis, 90 degrees
  if rotating about a or c).

    The phi value (PHIZONE) which places the unique symmetry axis in
  the plane normal to the X-ray beam and containing the rotation axis
  is determined, and a reflection list corresponding to a total
  rotation of PHITOT starting at phi=PHIZONE is generated.

   This reflection list is then compared to a list of all unique
  reflections for this spacegroup and the completeness and
  multiplicity is calculated, both as a function of rotation and
  resolution.

    It is assumed that all possible reflections are measured (ie none
  are lost because of spatial overlaps or because they extend over too
  many images). However, some reflections may be unobserved because
  they lie in the cusp region. The percentage of reflections within
  the cusp will depend on the wavelength, crystal symmetry and crystal
  orientation, and can be minimised by trying to orient the crystal so
  that the crystal axis closest to the rotation axis is at least
  THETAMAX degrees AWAY from the rotation axis, where THETAMAX is the
  maximum Bragg angle.

     It is often possible to collect data with a very high percentage
  completeness with a total rotation significantly less than
  PHITOT. This will inevitably result in a lowering of the overall
  multiplicity, but if data collection time is limited (for example at
  a synchrotron source) it is preferable to obtain a dataset with high
  completeness and less than optimal multiplicity rather than an
  incompete dataset with higher multiplicity ! 

     If the total rotation angle to be collected is specified, and the
  number (up to 3) of discontinous segments to be used, the program
  will determine the start and end phi values for each segment that
  will give the highest possible completeness. For example, a total
  rotation of 60 degrees in 2 segments for an orthorhombic spacegroup
  will result in the identification of two 30 degree segments which
  give the highest completeness.

      If some data has already been collected from one (or more)
  previous crystals, the program will determine the starting phi value
  for the "current" crystal that will give the maximum completeness,
  with the assumption that the phi rotation for this crystal is such
  that the TOTAL rotation for ALL the crystals is PHITOT (This assumes
  that all crystals are mounted about the same axis). The user may
  also define the rotation angle for the current crystal.


  Trying other phi ranges
  =======================

  After the initial calculation of completeness, the program will give
  the prompt:
P
  STRATEGY ==>
p
  Typing STATS at this prompt will give full statistics on completeness
  and multiplicity as a function of rotation angle and resolution.

  The effect of using other total rotations and different numbers of
  segments can also be tested (but using more than 3 segments is very
  time consuming)
P
  eg
  (AUTO) ROTATE 40 SEGMENTS 2
  RUN
p
  will find the best two segments of 20 degrees each for maximum
  completeness (the AUTO keyword can be omitted).

  Alternatively, the completeness of specified segments can be tested:
P
  START 0 END 20
  START 65 END 90
  RUN
p
     Note that the phi ranges specified on the START and END keywords MUST
  lie within the phi range generated by the program when it first
  starts. Thus if, for example, the keywords STRATEGY AUTO ROTATE 60
  were specified, then it is not possible to try: 
   AUTO ROTATE 90 SEGMENTS 1 RUN
  at the STRATEGY prompt (The program will complain).


  Use keyword EXIT or END to end the strategy option.

2 Examples

  Example command files
  ======================
  (i)   This example is to determine the best starting phi and phi
  rotation to collect data from an orthorhombic crystal, whose
  orientation is defined by the UMATRIX and MISSET keywords. More
  usually a MATRIX file will be supplied, as written by REFIX.
P
  STRATEGY SPEEDUP 20
  DISTANCE 80
  DETECTOR SMALLMAR
  UMATRIX 1 0 0 0 1 0 0 0 1
  CELL 50 50 80 90 90 90
  MISSET 12.0 13.4 15.0
  SYMM 19
  BEAM  90 90
  DIVERGENCE 0.35 0.3
  MOSAIC 0.2
  SEPARATION 1.5 1.5
  HKLOUT strategy.mtz
  POLARISATION MONOCHROMATOR
  WAVELENGTH  1.5418
  RUN
p  
  
  (ii) Example of the case when data have already been collected from one
  =======================================================================
  crystal
  =======

  
  See the User Guide for further examples.

  In this case data have been collected from phi = -20 to 19 from one
  crystal with missetting angles 12.0,13.4,15.0.
  
  A second crystal is mounted, which has missets of 7 23 -40. This run will
  determine the best phi range for this second crystal to complete the data.
  
  Normally the first crystal will have been collected starting at a
  zone. If this is NOT the case, it will probably be necessary to
  collect two segments of data from the second crystal to get complete
  data. This can be done by specifying "STRATEGY AUTO SEGMENTS 2" for
  the second crystal, and it may be advantageous to specify the sizes
  of the two segments. Thus if the first crystal was collected
  starting at 15 degrees away from a zone, for a total of 35 degrees,
  then the second crystal will need one segment of 15 degrees and
  another of 40 degrees (90-35-15) to get best completeness.
  
  (Normally the UMAT option should be used in REFIX in order to get
  the MISSETS of the second crystal relative to those of the first.)
P  
  STRATEGY  start -20 end 19 PARTS 2
  DISTANCE 80
  DETECTOR SMALLMAR
  UMATRIX 1 0 0 0 1 0 0 0 1
  CELL 50 40 73 90 90 90
  MISSET 12.0 13.4 15.0
  SYMM 19
  BEAM  90 90
  DIVERGENCE 0.35 0.3
  MOSAIC 0.2
  SEPARATION 1.5 1.5
  HKLOUT strategy.mtz
  POLARISATION MONOCHROMATOR
  WAVELENGTH  1.5418
  GO
  STRATEGY AUTO
  MISSET 7 23 -40
  GO
p  
1 SYMMETRY
    SYMMETRY Spacegroup name OR spacegroup number
P
  eg SYMMETRY R32  or SYMMETRY 155
p
        The crystal spacegroup, either as the name or the number in
    International Tables. Note that axial systematic absences (eg ok0
    with k odd in spacegroup P21) are measured by MOSFLM, so that the
    symmetry can be checked. Lattice absences however (due to face or
    body centring) will NOT be measured.

    For rhombohedral space groups (eg R3, R32) the equivalent
    hexagonal setting will be used by default. To use the rhombohedral
    cell for indexing, use the LATTICE keyword. See the LATTICE
    keyword for more details.

1 SYNCHROTRON
  SYNCHROTRON Keywords
        Parameters that must be specified are:-
P
                WAVE   - Wavelength - Set explicity on separate card.

                DIVH   - Horizontal divergence of source.

                DIVV   - Vertical divergence of source.

        *****   DISPERSION - Wavelength dispersion - Set explicity on
                             separate card.

                DELCOR - Correlated component of wavelength dispersion.

        *****   MOSAIC - Mosaic spread - Set explicity on separate card.

                POLARISATION - The polarisation of the beam. Default
                               value is 0.95 which is appropriate for
                               the SRS.
p
         The following parameters can be set explicity by using the keywords:
P
                DIVH x
                DIVV x
                DELCOR x
                POLARISATION x
p
        in any order following the 'SYNCH' keyword and using any number
        of cards. The other must be set explicity by appropriate cards.

        OR implicity by using keywords
P
                PX72 or PX96 
p
        which will set the parameters DIVH,DIVV,DISP,DELCOR to standard
        values for PX stations for 1 dot blue collimator (0.3mm.) which are
        defined in a DATA statement.


    ***********************************************************************

        Following tests conducted January 1988, for data collected using
        the one dot blue collimator at Daresbury (post HBL), the following 
        values are set up as defaults:
P
        Station 7.2     Horizontal divergence = 0.15 degrees
                        Vertical divergence   = 0.05
                        Dispersion            = 0.0015
                        Correlated dispersion = 0.0

        Station 9.6     Horizontal divergence = 0.06 degrees
                        Vertical divergence   = 0.02
                        Dispersion            = 0.0015
                        Correlated dispersion = 0.0
p
        These values, when used in conjunction with a mosaic spread of
        0.05, give good agreement with observed partial fractions in
        POSTCHK. Some crystals may well have a larger mosaic spread,
        and POSTCHK should always be used to validate the value
        assumed for mosaic spread.

1 TEMPLATE

  TEMPLATE <string>

  The TEMPLATE allows a completely general way to define image
  filenames. The general form of the "string" is:
P
  "string1"###"string2"

  eg fred###ab.raw
  or: myfile#####
  or: xr2_3_##.zo
p
  The number of # symbols given MUST be equal to the number of digits
  to be interpreted in the image filename.

  Thus TEMPLATE fred_##.img  would be able to read images
  "fred_01.image" up to "fred_99.img" but would not cope with a three
  digit image number such as "fred_100.img".

  However, TEMPLATE fred_0###.img will read images "fred_0001.img" to
  "fred_0999.img", but not "fred_1000.img".

  Note that the image filename does not need to have an extension.

  If a TEMPLATE and an IMAGE keywords are given, the TEMPLATE keyword
  MUST come first.

  A different TEMPLATE keyword can be given for a jobwith multiple
  RUNs, as with the IDENT keyword.



1 TESTGEN

  The TESTGEN option will calculate the maximum allowed oscillation
  angles as a function of the phi value for a given maximum acceptable
  percentage of overlapped reflections (which can be zero).

  It will most commonly be ussed in conjunction with the STRATEGY
  option (qv).

  TESTGEN START phstart END phend STEP phstep MINOSC rotmin MAXOSC rotmax
          OVERLAP x

        START phstart
            Define the starting phi. This keyword MUST be given.

        END phend
            Define the ending phi. This keyword MUST be given.

        STEP phstep
            The optimum rotation angle will be calculated every
            "phstep" degrees between "phstart" and "phend".
            Default: 5 degrees

        OVERLAP x
            The maximum rotation angle giving less than x% overlapped
            reflections will be calculated. Note that x is in PERCENT.
            Default 0%

        MINOSC rotmin
        MAXOSC rotmax
            Only rotation angles between "rotmin" and "rotmax" will be
            considered.
            Default: rotmin 0.2, rotmax 5.0

        ANGLE oscang
            If the ANGLE keyword is given, then the overlap for a
            fixed oscillation angle "oscang" is calculated between
            phi=phstart and phi=phend. No attempt to find the "best"
            oscillation angle is made.

  Note that the TESTGEN keyword can be given at the STRATEGY prompt or
  at the normnal MOSFLM prompt (without running the STRATEGY option).
  
2 More

    Allows an optimum oscillation angle to be determined to avoid an
  excessive number of spatial overlaps. 

    The program will generate reflection lists for all angles between
  phstart and phend in steps of phstep. For each angle, the MAXIMUM
  oscillation angle which gives fewer than "x" overlaps (where
  x is a PERCENTAGE of the total number of reflections generated)
  will be determined, using oscillations between "oscmin" and
  "oscmax".
  
  The completeness analysis in the STRATEGY option assumes that NO
  reflections are spatially overlapped. Providing that spots within a
  lune (ie in the same plane in reciprocal space) are not overlapping,
  spatial overlaps can usually be reduced to an acceptable level by an
  appropriate choice of oscillation angle for each image.
  
  The determination of whether or not a reflection is spatially
  overlapped depends crucially on the mosaic spread, beam divergence
  parameters and the minimum allowed spot separation. The mosaic spread
  and the minimum spot separation can be reset at the STRATEGY prompt to
  test how critical these values are, using keywords MOSAIC and
  SEPARATION respectively.
P  
  eg TESTGEN START 0 END 90 OVER 1
p
   Will determine the MAXIMUM oscillation angle between phi=0 and
  phi=90 (in steps of 5 degrees) that give less than 1% overlaps.

1 THICKNESS

   THICKNESS <THICK>

   Sets the nominal thickness (in mm) of the detector.  This affects
  the the expansion of the peak component of the measurement box to
  allow for obliquity and so will determine the overall size of the
  measurement box of the standard profiles. Note that the background
  rim parameters are optimised individually for each standard profile.


   Default: Image plate 0.2mm


1 TIMEOUT
   TIMEOUT <x>

   Sets the time (in seconds) for the timeout when running
   interactively with the X-window graphics and the Timeout mode set.

   Default: 1 seconds

1 TITLE
   TITLE string
        A string of up to 80 characters. This will be stored in the
    generate file and transferred to the MTZ file.


1 TWOTHETA 
   TWOTHETA <x>

   For a detector with a two-theta arm, this specifies the rotation
  angle in degrees for the two-theta arm. A positive rotation is
  counter-clockwise when viewed looking down the rotation axis (from
  the crystal towards the goniometer). (This is the convention adopted
  for R-axis scanners). Note that the direct beam position (keyword
  BEAM) is assumed to be at two-theta=0 unless the subkeyword
  SWUNG_OUT is included with the BEAM keyword (qv). 

  For a swung out detector the areas for the standard profiles MUST be
  explicitly defined using the PROFILE XLINES YLINES keywords (see
  under PROFILE).

  Default 0.0

1 UNPACK
   UNPACK <img1> <img2>

   This allows a compressed Mar345 image file to be converted to a
   standard "IMAGE" type file. However, all pixel values greater than
   65535 in the original image will be set to 65535. The overload
   records in the original file are not used, and no overload records
   are generated in the new file. Although a standard length header
   record is written to the output file, it ONLY contains the image
   size (in the first two 4-byte words), all other header information
   is lost. The structure of the new file will then be:
   A 4096 byte header record.
   XPIX records each of length YPIX, with pixel values stored in
   integer*2 variables, range 0-65535.
   The name of the new file will be the same as the original file but
   with the filename extension ".image". Thus fred_001.mar2300 will
   become fred_001.image.
   img1 and img2 are the numbers of the first and last images to be
   converted.
   An example input would be:
P
   IDENT fred
   EXTENSION mar2300
   UNPACK 1 to 20
   GO
p
1 UMATRIX
   UMATRIX <x x x x x x x x x> OR filename

        The U matrix specifying the crystal orientation. If this
    matrix is given, then the cell dimensions and the wavelength must
    also be specified.

        If a filename is given here, then the [U] matrix, cell
    parameters and missetting angles will be read from that file.
    This file can be output from IDXREF by using the OUTFILE keyword
    or from REFIX (written to logical filename MATRIX).  If a
    subsequent CELL or MISSET keyword is given, the values read from
    the file will be superceeded. Similarly, the values read from the
    file will superceed values given on earlier CELL or MISSET
    keywords. A warning message is given in each case.  (*** Note that
    this file has a total of 9 records. The first 4 contain the [A]
    matrix and missetting angles, followed by the [U] matrix, cell
    parameters and missetting angles. ***)


1 WAIT
  WAIT <x> <y> <z>

   If a WAIT keyword is given, and an image file that is to be
   processed does not exist, the program will wait for <x> minutes 
   and <y> seconds, testing at short intervals (maximum 5 seconds) 
   whether the file exists. It will then proceed normally if the file 
   does exist. For compatibility with previous versions of Mosflm, if 
   the file does not exist, it will continue to wait for one further 
   period of <x> minutes and <y> seconds while trying again. If the 
   file still does not exist, the program will stop.

   The third parameter, <z> (in seconds), has been introduced to provide
   an extra short delay to ensure that the image has been completely 
   written. Its use should be unnecessary for CCDs, but may be useful for
   image plate detectors.

   This feature has been introduced to allow images to be integrated
   as the data is being collected. The delay time <x> <y> should 
   therefore be set to at least the exposure time plus the readout time. 
   
   As the program now tests for the file's existence regularly, it is safe
   to use the following example input if there is an irregular delay in 
   images being written to disk.
P
   WAIT 60 0
p
   will wait for a maximum of 120 minutes for the next image, but will 
   check every few seconds.

   To integrate and scale/merge data as it is being collected, use the
   HKLOUT MULTIPLE option...see HKLOUT keyword.


1 WAVELENGTH
  WAVELENGTH <x>
        Wavelength in Angstrom. (default is CuKalpha=1.5418)

        Default: 1.542, but for Mar Research, Raxis and Mac Science
    scanners the wavelength will be read from the image headers.

1 XGUI
  XGUI [ ON | OFF ]
  This option is not really intended for users, but can be manipulated
  to write out a uuencoded JPEG of the current image using a set of 
  sub-keywords.

  subkeyword CREATE_IMAGE [ BINARY TRUE     | PREDICTION TRUE  ]
                          [ QUALITY <x%>    | ZOOM <factor>    ]
                          [ WIDTH <width>   | HEIGHT <height>  ]
                          [ XMIN <xmin>     | XMAX <xmax>      ]
                          [ YMIN <ymin>     | YMAX <ymax>      ]
         
2 BINARY TRUE
  Write a binary JPEG file rather than uuencoding it. This is not the default.

2 PREDICTION TRUE
  If a UB matrix, wavelength, mosaic spread, beam center and distance have been
  specified, overlay a prediction over the image.

2 QUALITY <x%>
  Produce the JPEG file with quality <x%>, default 75%.

2 ZOOM <factor>
  Zoom the image by the specified factor 

2 WIDTH/HEIGHT
  Specify the dimensions of the output image file in pixels.

2 XMIN/XMAX/YMIN/YMAX
  Specify the area of the original image to be written to the JPEG file.

  Examples: 
  
  XGUI ON                                         !
  GO                                              !
  CREATE_IMAGE BINARY TRUE FILENAME <filename>    ! writes a binary JPEG file 
  RETURN                                          ! called <filename>
  EXIT                                            !

  XGUI ON                                          !
  GO                                               !
  CREATE_IMAGE PREDICTION TRUE FILENAME <filename> ! writes a uuencoded JPEG 
  RETURN                                           ! file called  <filename>
  EXIT                                             !

  The commands can be concatenated.

1 Documentation
2 Measurement_box
3 Description
  The Measurement Box

   The measurement box is a rectangular  array  of  pixel
  values  covering each spot on the image, allowing for estimation of the
  integrated intensity of each spot:  thus it must
  include an area of background around each spot as well as the whole of
  the spot.  The edges of this box are parallel to the Xd and Yd axes of
  the detector.
P
             <------------------ NXS = 23 --------------->

        ^    - - - - - - - - - - - - - - - - - - - - - - -  ^
        !    - - - - - - - - - - - - - - - - - - - - - - -    NRY =2
        !    - - - - - -                       - - - - - -  ^
        !    - - - - -                           - - - - -
        !    - - - -                               - - - -
        !    - - -                                   - - -
        !    - -                                       - -
        !    - -                                       - -
     NYS =17 - -                                       - -
        !    - -                                       - -  ^
        !    - -                                       - -  !
        !    - - -                                   - - -  !
        !    - - - -                               - - -    !
        !    - - - - -                           - - - - -  NC =8
        !    - - - - - -                       - - - - - -  !
        !    - - - - - - - - - - - - - - - - - - - - - - -  !
        ^    - - - - - - - - - - - - - - - - - - - - - - -  ^
            <NRX> =2
p

   Measurement box parameters.

   NXS, NYS (odd integers) define the overall dimensions of the
  rectangular array of pixels for each spot.
   NRX, NRY are the number of columns or rows of points in the
  background rim and NC the corner background cut-off which 
  corresponds to a half-square of side NC points.

   If NC=0 the pixels in the rectangular corner defined by 
  NRX,NRY are omitted from peak and background.

   To take account of the variation of spot shape  and  size  across
  the diffraction pattern the measurement box parameters are defined for
  spots at the centre of the detector and automatically  expanded  to  
  allow for  oblique incidence and the finite detector thickness 
  (see Expansion).
  
   The origin for expansion is the detector centre,Only the central
  peak area is expanded; the number of points included in the
  background will increase correspondingly if a rim is defined for a
  measurement box, but not for corners alone.


3 Optimisation

   Choice of Box Parameters. The measurement box parameters are
   optimised automatically by the program (see PROFILE PROPT for 
   further details).
  

3 Expansion
   Measurement Box Expansion

   The portion of the measurement box corresponding to the
  peak  is  automatically  expanded to allow for variations in
  spot size and shape over the image.  The diffracted  beam  is
  considered  as  a  parallel  pencil  of X-rays, no allowance
  being made for its divergence with increasing distance  from
  the  crystal.   The  actual  spot  expansion  is in a radial
  direction but this is resolved into components along the Xd,
  Yd directions so that the peak area remains rectangular when
  expanded and for a given Xd co-ordinate, all boxes will have
  the  same number of points along Xd and likewise for Yd.  If
  w is the width of the beam for normal incidence then for  an
  angle of incidence beta the dimensions on the image becomes

       w' = w*sec(beta) + t*tan(beta)

    where tan(beta) = Rf/D and t is image thickness.

   If the angle of incidence is resolved into components along Xf and Yf 
  we have

        tan(betax) = Xf/D  ;  tan(betay) = Yf/D

   and to a first approximation

        sec(betax) = 1 + (Xf)**2/2*(D)**2

   Thus

        wx' = wx*( 1 + (Xf)**2/2*(D)**2 + t*Xf/D ........... (1)

   and similarly for wy'.

   Modification to the raster expansion

   The optimum dimensions of the measurement box and the definition of
  peak and background areas is normally done automatically (PROFILE
  PROPT) using the average spot profile for the centre of the image as
  produced by CENTRS.  However, this profile includes reflections
  which are up to 45mm ( controlled by keyword LIMIT) in X and Y from
  the direct beam position centre, which will therefore be larger than
  the spot size at the centre of the image.  This will therefore
  give an overestimate of the spot dimensions wx, wy which will lead
  to over-expansion of the measurement box on outer regions of the
  image.

   The spot dimensions (ie peak dimensions) chosen on  the
  basis of the average spot profile are therefore corrected to
  give a true estimate of the spot size at the image centre  by
  re-arranging equation (1) to give:

     wx = (wx'-t*Xf/D) / (1+Xf**2/(2*D**2))

  and similarly for wy.  Here wx' is the peak width chosen for
  the average spot profile (ie (NXS-2*NRX-2)) and Xf is set by
  default to be 20mm, although this can be changed by  keyword
  'CENTRE'.   This  value  of wx rather than the peak width of
  the average spot profile  is  then  used  to  calculate  the
  expansion  of  the  peak dimensions using equation (1).  (It
  should be noted that  the  measurement  box  is  never  made
  smaller  than the dimension used to display the average spot
  profile.)

   In practice this leads to a  significant  reduction  in
  the  total  number of expansions applied, and consequently a
  reduction in the number of standard profiles that have to be
  determined.

3 Calculations
   Calculations done on measurement box
  
  Plane fitting to background

  We define a plane

     ap + bq + c = 0

  with  origin  of  coordinate  system  (p,q)  at  centre   of
  measurement  box  and  axis  parallel  to edges of box.  The
  least-squares best plane is obtained by minimising

    PHI = (sum over B)[( ap + bq + c - D )**2]

   where the sum is taken over all background points, (B) and D
   is  the  measured  optical  density.  Because the background
   points are symmetrically disposed (ie (sum p, sum q, sum p*q
   all zero) the best plane is given by minimising

      a = (sum over B)[p*D]/(sum over B) [p**2] 

      b = (sum over B)[q*D]/(sum over B) [q**2]

      c = (sum over B) [D/N(B)]


   When background points have been  rejected (either because of
  intruding neighbouring spots (calculated by the program) or
  outliers from the best least squares plane using BGSIG)
  the background area will no longer have mm symmetry and  the
  equations  above  are  no  longer  valid.  In such cases the
  equations given by Rossmann (1979) are used.

   Centre of gravity

   The centre of gravity of the spot (S) relative  to  the
  centre of the box after removing the background contribution
  is given by

    delta(x) = [ (sum over S) [p*D] - a*(sum over S) [p**2]]/I 

    delta(y) = [ (sum over S) [q*D] - b*(sum over S) [q**2]]/I

    where the sums are taken over the spot area (S) and  I,  the
    background subtracted integrated intensity, is given by

             I = (sum over S)[ (D) - C*N(S) ]


  References
P
   Rossmann, M.G. (1979) J. Appl. Cryst 12, 225-238.
p



2 Input_files
   Input Files

   The program requires only one input data file: an image file.

  At present the program will read Mar Research, R-axis (II or IV),
  Mac Science, Molecular Dynamics, Fuji, ESRF CCD and LMB packed images.

   The filename is set up automatically as  the  crystal
  identifier  (as  read  from the generate file) followed by the 3 digit
  pack number (right justified and padded with zeros) suffixed by  the
  default extension (can be specified by keyword EXTENSION)
  and prefixed by DIRECTORY (keyword DIRECTORY) which should specify the
  disk and directory eg /scr0/andrew/ where the image is stored.
P
  eg. directory:          /scr0/andrew/
      crystal identifier: catx96
      image:              3
      od filename:        /scr0/andrew/catx96_003.image
p
  Additional files
   a) PROFILES
   If a set of profiles have been written to a file using the
  SAVE subkeyword on PROFILE card, then these can be read and used to
  measure a subsequent image using PROFILE READ. The profiles are read
  from logical file PROFILE. A template should be assigned to this
  logical, from which the filenames for the profiles of successive
  blocks of images will be generated.
P
  eg. assign/user catx1 profile
p
  will result in profile filenames:
P
  catx1_001.prf    for the first block
  catx1_002.prf    for the second block
  etc etc
p
   b) DUMPSPOTS
   A list of reflection indices may be given in a file assigned
  to logical SPOTDUMP and a large amount of debug information on the
  integration of these reflections will be generated. A maximum of 50
  reflections is allowed.
  See also keyword BAD for further information.

2 Output_Files 
  Output files from MOSFLM.
   a) PROFILES
  If the SAVE option has been included on the PROFILE card, then
  the standard profiles will be written to logical file PROFILE
  using automatically generated filenames (see Input_Files).

3 Temporary_Files
  TEMPORARY FILES USED BY `MOSFLM'

  a)SPOTOD (Unit 3)
     Created  by  `MEAS'.   Used  to  store  the pixel values
     for every spot measured.  The file is read by `PROCESS'. 

     It is written to logical file SPOTOD, which can be ASSIGNED by 
     the user. The file may be large, so it should always be deleted 
     at the end of the job. From Version 6.2.3, it is written as a 
     scratch file, and is automatically deleted at the end of the job.
     

   Format is as follows:
      First record contains: XCEN,YCEN,NREF-NPARTEND,MAXR,NPARTEND
      Then subsequent records are:
      Word 1 Record number of reflexions in  the  `generate'  file
      (made negative for partially recorded reflexions).

      Word 2 X co-ordinate in detector units.

      Word 3 Y co-ordinate in detector units.
                                      
      Word 4 NXX,  NYY  expanded  measurement  box  dimensions  (2
      bytes).

      Word 5 Profile box number

      Words 6-9 Not yet used

      Word 10 onwards:  pixel values for the  measurement  box
      The pixel values are I*2 words (NXX x NYY).

           This set of records is repeated for each reflection.

   b)BADSPOT.DAT (Unit 46) 
      Created by PROCESS or SPROCESS only when
      running  online, contains optical densities of 'badspots' or
      spots specified in SPOTDUMP.DAT (see keyword BAD).

2 Residual_vectors
  Residual systematic errors
       The plotting of residual centre of gravity shift vectors for  the
  fully  recorded  spots  in  `PROCESS'  will  often indicate systematic
  errors which have not been accounted for  by  the  refinement  of  the
  film/drum  transformation.   A  value of Yscal significantly different
  from 0.998 can be indicative of systematic errors in cell  dimensions.
  The principal causes of residual errors are:

   1.  Incorrect orientation

       Since spot positions are independent of phiz and psix (see IDXREF
  for definitions) only an error in psiy can give rise to residual shift
  errors.  The plotted vector shift pattern will show  shifts  in  spots
  close  to the rotation axis.  The shift directions are opposed on each
  side of the rotation axis (mirror plane) and of opposite sign  at  top
  and bottom of the film.

   2.  Relative cell parameters in error

       Will usually give rise to a pattern of vector  shifts  in  which,
  along  the  Xf  direction,  the vectors are all outwards from the film
  centre, while along Yf they are all inwards.



3 Instrument_errors                 
    The standard deviations for the intensities described under Errors
  allow only for random errors due to granularity of the film. However
  there will also be a significant contribution due to errors in
  measuring the optical densities. There are at least three sources of
  error:
   1) The limited slew rate of the photometrics means that
  very rapidly changing optical densities will not be followed
  correctly.
   2) Positional errors in positioning the photometrics will
  mean that the true optical density is not measured correctly.
   3) Possible flare around strong spots.
    All of these errors will be more serious for strong reflections,
  where the optical density gradient is greater. In order to make some
  allowance for this source of error, a very simple-minded model has been
  adopted.
    It is assumed that the error is primarily due to positional errors
  along the detector rotation axis, and the error is therefore proportional
  to the optical density gradient along the detector X axis. The constant
  of proportionality is called the "detector error factor" KSC. Thus:

   sigma(od) = KSC * d(od)/dx

    It is further assumed that reflection profiles are triangular in cross
  section along X and Y, so that the optical density gradient is constant
  at all pixels within the peak for a given reflection. With this assumption
  it can be shown that the total integrated intensity of the reflection is
  given by:

   I = ((NX+1)/(NY+1))**2 x (NY**3+3*NY**2+5*NY+3)/12 x d(od)/dx

  where NX,NY are the full widths of the reflection (in pixels) in the X
  and Y directions, and d(od)/dx is the optical density gradient along X.
  Therefore, with the assumptions listed above, and knowing the reflection
  dimensions, d(od)/dx can be calculated from the integrated intensity.

    The contribution of this error to the total variance of the intensity
  is then

   ERRTERM = NPK * (KSC*d(od)/dx)**2

  where NPK is the number of pixels in the peak. 

    In order to obtain a value for KSC, the goodness of fit of individual
  reflection profiles to the standard profiles is used. The expected
  goodness of fit due to the film granularity can be calculated (see Errors)
  and this is compared to the observed fit. The difference is assumed to be 
  due to detector instrument errors. Thus PKRATIO is defined as the ratio
  of the observed rms fit of the reflection to the standard profile
  to the expected fit based on the film granularity plus the detector
  contribution (whose variance is ERRTERM above). A value for KSC is
  calculated which makes this ratio unity for all reflections (it
  actually minimises the deviation from unity in a least squares sense).
    The success of the model can to some extent be judged by the variation
  of the PKRATIO with intensity; it should be 1.0 for all intensity ranges.
    It must be realised, however, that KSC is derived from the fit of 
  individual reflections to the standard profiles. Thus if there is any
  error in the standard profiles, this will show up as a larger value
  for KSC. In particular, if the individual spots show "fine structure",
  as is very commonly observed for films recorded at Daresbury SRS Wiggler
  station post HBL, because the main beam has more than one component,
  or if the crystal is split or otherwise damaged, then this too will
  show up in the value of KSC. Consequently, KSC will not be a strict
  constant for a particular detector, but will show some variation depending
  on the quality of the films.
    In the limited experience obtained so far at Imperial, values of
  KSC between 0.3 and 0.5 have been observed for quite different films
  in terms of exposure, spot size and resolution. The values are remarkably
  constant for a given dataset (from one crystal) varying by less than
   0.1. For reasons not yet understood, KSC is systematically slightly
  greater for B and C films than A films. Given the assumptions on which
  KSC has been derived, the observed variation is quite acceptable.
    Within one dataset, KSC can be used as an indicator of whether a
  film has been measured correctly, and its value is written to the ouput
  and the summary file.
    Because of the way KSC is derived, in cannot be calculated unless
  profile fitting is used. However a value can be supplied (keyword
  DETECTOR) which will always be used (even if profile fitting is performed).

    The magnitude of KSC (say 0.3) suggests a sigma(od) value of about
   10 grey levels for a very strong reflection, which is significantly 
  greater than observed values obtained by making repeated scans across 
  a reflection (this gives sigma values of between 4 and 5 at the points
  where the gradient is greatest, and between 1 and 2 over most of the
  reflection). In view of the simple model this is perhaps not too
  suprising. 



1 Output

       When running online, a copy of the terminal ouput is  written  to
  UNIT 8 which may be printed as a record if desired.  In batch mode the
  output is directed to UNIT 6.

  BEWARE:  DEBUG output produced by the  DEBUG  keyword  is  not  always
  written  to  UNIT  8  when running online, and is sometimes written to
  UNIT 8 rather then UNIT 6 when running in batch mode, so check both !

       Most of the output is self-explanatory but  some  points  require
  clarification.

       All keyworded input is reflected, and if no keyword is given  the
  default  values  for  important  variables  used  in  profile  will be
  printed.

       FULLY RECORDED REFLECTIONS TO BE MEASURED refers to the  list  of
  reflexions in the central area of the film.

       The STARTING RESIDUAL is the root mean square difference  between
  the  observed  c of g and calculated spot positions before refinement.
  If the starting residual is greater than a preset limit  (80  microns)
  the  measurement  box was not sufficiently well centered and the cycle
  of refinement is repeated.  The final rms residual (in mm) and a 
  weighted residual are printed together with  the refined
  transformation parameters. The weighted residual should be close to
  unity independent of the strength of the image. The positional
  residual (unweighted) will in general be larger for weaker images.
  A final residual greater
  than a preset value (defaults to 60 microns, can be changed by keyword
  'RESIDUAL')  causes  the program to enter the film-plotting subroutine
  if running on-line or go onto the next pack if running in batch.   For
  most  films  the  residual  at this point in the program is reduced to
  0.01mm: much larger values would indicate a  systematic  error
  in  calculated  co-ordinates  due  to  orientation  or  cell parameter
  inaccuracies.  In the majority of cases TILT and TWIST are  less  than
   20 (1/100th  degree  units)  while  BULGE is commonly around 50.  The
  refined parameters D1,D2,PHI and PSI for precession films are  defined
  in section 5 above.

       Optimum raster box parameters are determined for the average
  spot profile for spots in the central region of the detector for the
  first image in each block of images (PROFILE PROPT). Note that the
  parameters for each of the standard profiles are determined
  independently.

       For the average-spot profile the c of  g  shifts  are  in mm
  units  and  the  background  residual  is  the  rms  deviation  of the
  background optical densities from the best plane (with the peak
  scaled to a vlaue of 255 arbitrary units).   The
  average  spot-profile,  which is scaled internally in the range 0-255,
  is printed using the 36 characters 0-9, A-Z.   Negative  signs  prefix
  all  background  points and * prefixes omitted points.  The profile is
  calculated following the final cycle of  refinement  by  CENTRS.
  If running online the measurement  box parameters can be updated at
  this point.

       Refinement is then carried out using selected reflections over
  the whole detector (rather than just the central region).

       If post-refinement has been requested, the results of the 
  post-refinement will be given at this point. The initial and refined
  values of parameters and their estimated standard deviations will be
  given.

       If any parameter has shifted by more than a specified amout,
  the image will be remeasured using the updated parameters.




  STATISTICS ON THE STANDARD PROFILES (BEFORE AVERAGING)

          The parameters listed for each profile are:

  BOX
            The profile box number.

  XMIN,XMAX,YMIN,YMAX
            The limits (in mm with respect to the lower left of the
            image viewed from behind image towards the source)  of  the
            rectangular  area  of  the  film  where this profile will be
            applied.  X and Y are in the detector coordinate frame.

  NX,NY
            The X and Y dimensions of the  measurement  box  (in  raster
            units).

  RMSBG
            The r.m.s.  variation in the  background  pixel values
            after  background point rejection (based on calculated
            positions of neighbouring spots and using the BGSIG
            parameter, althoug this should normally be set so high
            that all rejected pixels are due to intruding neighbouring
            spots. This value is on a scale where the maximum profile
            value is 255.


  DELX,DELY
            The deviation of the c.  of g.  of the standard profile from
            the centre of the measurement box (10micron units).

  NO
            The number of reflections included in the profile.

  A,B,C
            The average background  plane  constants.   'A'  gives  the
            gradient  parallel  to  detector  'X',  'B'  that parallel to
            detector 'Y' and C is the constant.  A and B are in counts
            per pixel and C is in counts. Thus C represents the
            average value of the background in that region of the
            detector.



  DEVMAX
            The largest deviation of a background point  from  the  best
            least-squares  plane after background point rejection.  (See
            note below.)

  NBGREJ
            The number of background points rejected by the BGSIG test.

  NOVER
            The number  of  overloaded  reflections  rejected  from  the
            profile.

  NWK
            The number of reflections too weak to  be  included  in  the
            profile.
  CORRLN
            The correlation coefficient between the profile for the
            central region of the image and that part of this profile
            which corresponds to the UNEXPANDED peak area.  Typical
            values range from 0.8 to 1.0.




       STATISTICS ON AVERAGED PROFILES.  These statistics are  identical
  to  those  given  above,  with  the exception that the backgrounds are
  subtracted from individual profiles prior to averaging (because  there
  may  be  significant  differences  in  the  mean  background  even for
  neighbouring profiles).  Thus the background plane constants should be
  close to zero.

       PRINTED PROFILES.  The standard profiles  are  displayed  in  the
  same  format  as  the average spot profile for the centre of the image.
  Background points rejected from  the  profiles due to the intrusion
  of neighbouring spots or by the BGSIG test are preceeded by a '*'.  
  Note that at this stage all profiles have been background-subtracted
  and placed on a common scale.
  All negative profile values are represented by ']'.

       LIST OF BADSPOTS.  Reflections failing  the  tests  on  INTENSITY
  (negative with magnitude more than 5*sigma),BGRATIO (greater than 3.0)
  PKRATIO (greater than 3.5) or too high a gradient are listed.
  'H K L' are the  reflection indices.
  'M'  is  zero for fully recorded reflections or 100*fraction
  recorded for partials.  The sign of 'M'  depends  on  whether  it  was
  recorded at the beginning or end of the oscillation range.
  'XG' and 'YG' are the GENERATE file coordinates in 10micron units,
  XS and YS are the detector coordinates in pixels.
  'INT'  and  'SD' are the integrated intensity and standard deviation.
  The values for BGRATIO and PKRATIO are followed by the background
  plane constants for this spot.

       DETECTOR ERROR FACTOR. This gives the factor used to derive
  the contribution of detector (instrument) errors to the standard
  deviation of the measured intensities. See Documentation, Errors 
  (Instrument) for a full description.


       The total number of reflections measured, and a breakdown into
  fulls, and overloads is given.  The variation in BGRATIO, PKRATIO
  and standard deviation with intensity is given (The summation
  integration intensities are used for the binning). <SDI> and <SDP>
  are the mean summation integration and profile fitted standard
  deviation in the intensities respectively.  BGRATIO and PKRATIO are
  defined in Documentation, Errors and both should have values close
  to 1.0 for all intensity ranges.  A value of BGRATIO greater than
  one for all intensities suggests the gain is underestimated or the
  refinement is not satisfactory or the measurement box parameters
  require modification (the average spot profile can be used to check
  this).  Occasionally the value increases sharply for the strongest
  intensities indicating that the peak is tending to run into the
  background area.  If PKRATIO deviates significantly from 1.0 for
  lower intensities this suggests either the gain is incorrect (in
  this case the BGRATIO should also differ from 1.0 in the same way)
  or, if BGRATIO is 1.0, the standard profiles are inadequate for some
  reason.  If PKRATIO is unity for lower intensities but not for
  higher intensities this suggests that the instrumentation errors are
  not being modelled correctly.

       There follows an analysis of mean intensity and standard
  deviation as a function of resolution for profile fitted and 
  summation integration intensities. This is very useful in assessing
  the quality of the data.

       INTENSITY AS A FUNCTION OF PARTIALITY tabulates the mean
  summation integration intensity (and the standard deviation of that
  mean) as a function of the fraction recorded, using the "M" values
  stored on the generate file and converting these to the fraction
  recorded using a sinusoidal rocking curve model. The mean intensity
  should, obviously, increase as the fraction recorded increases, but
  due to the limited number of reflections in each bin, and the fact
  that reflections at higher resolution will be weaker anyway, some
  variability will be seen.  In particular, the mean intensity may
  very well NOT drop to zero in the 0-5% recorded range, even if the
  mosaicity was overestimated.  This arises because diffuse scatter at
  Bragg positions can be quite strong, particularly on images recorded
  at the synchrotron. The table should therefore be used with caution
  to estimate crystal mosaicity (try POSTREFINEMENT instead).

       The  Histogram of rejected points indicates how many
  reflections have a given number of background points rejected,
  either because of intruding neighbouring spots or because they are
  outliers (judged by the BGSIG parameters) from the best least
  squares plane.


       The R-FACTOR between integrated intensities  and  profile  fitted
  intensities  is  given  seperately  for  fully  recorded and partially
  recorded reflections.  Typical values are 3-9% for fully recorded  and
   6-12% for partials. If requested ('PRINT' keyword) a breakdown of the
  differences between the profile-fitted and integrated intensities  and
  their  standard  deviations  is given as a function of measurement box
  number and intensity. It is not possible to attach too much weight to the
  absolute values of these R-factors. R-factors of zero are NOT desirable
  because they imply that profile fitting is not making any difference.
  However, R-factors of 20% (except for very weak reflections) would imply
  larger differences in summation integration and profile fitted intensities
  than might be expected. The R-factor should, however, show two trends:
    1) It should always be larger for partially recorded reflections than
  for fully recorded ones.
    2) It should have very similar values for all images collected from one
  crystal. If one image shows a significantly larger R-factor it suggests
  some problem with processing.
   The R-factors are best used as indicators of problem images, rather than
  an absolute indication of the quality of the data as a whole.



       In the analysis of c.  of g.  shifts for fully recorded spots with
  an integrated intensity greater than 50, DX and  DY  are  with  respect  
  to the detector coordinate frame and are in 10 micron units.  Typical 
  shifts are between 10 and 40 microns.  A shift greater than 60 microns
  suggests  an  error  in cell parameters or that the refinement has not
  been successful.


1 Summary 

       A summary of the results of processing each image  is  written  to
  logical   filename   'SUMMARY'. 


  PACK
  The image number (as given on the PROCESS keyword

  CCX,CCY,CCOMEGA

  The camera constants (in mm and degrees). The magnitude of CCX and
  CCY reflects the accuracy of the supplied direct beam coordinates.
  However their actual magnitude is really immaterial (providing they
  do not exceed your expected error in the direct beam coordinates)
  but they should remain CONSTANT throughout the data collection
  (assuming the beam is not moving !). If using the post-refinement
  option to refine crystal missetting angles, then CCOMEGA may show a
  gradual drift, since any errors in orientation around the direct
  beam are taken up by CCOMEGA.

  DIST.

  The refined crystal to detector distance.

  YSCALE.

  The relative scale factor in the detector Y direction. This should be
  very close to unity (to within 1 part in 1000 or better) for
  detectors which have equal pixel sizes in X and Y (For R-axis II
  scanners, it should be 1.0324). Deviations from unity indicate
  errors in cell parameters.

  TILT,TWIST

  Deviations from normal incidence on the detector. TILT is a rotation
  about a horizontal axis, TWIST about a vertical axis, and the values
  given are expressed in hundredths of a degree. These values should be
  close to zero (less than 20) and should be CONSTANT (to within
  experimental error).

  ROFF,TOFF

  The radial and tangential offsets for the detector (see MOSFLM notes) in
  mm.  These should be CONSTANT to within experimental
  error.  Watch out particularly for variations in ROFF by one pixel.
    This indicates that the limit switch on the scanner is not
  functioning correctly (this should only occur very rarely if at all).

  RESID

  The rms positional residual (in mm) after refinement of the
  detector parameters. For strong images this should be between 0.02 and
  0.04. For weak images or large spots (due to high mosaic spread) it
  can be significantly higher. If partials have to be
  included in the refinement it will also be higher. A dataset with rms
  residuals of 0.2 to 0.3 can still give a final Rmerge of under 10% !
  Remember a residual of 0.15 is still only one pixel.  If the residual is
  greater then 0.04 for a strong image, there is almost certainly an
  error in the cell parameters, and they should be refined using the
  POSTREFINEMENT options. Even for weaker images, the positional
  residual for the initial refinement (using only central spots) should
  be small.

  WRESID

  The weighted residual. This should be close to unity (independent of
  the strength of the image). Larger values suggest errors in cell
  parameters or crystal orientation.

  FULL,PART,OVRL,NEG

  The number of fully recorded, partially recorded and overloaded
  reflections measured on the image (this includes reflections with part
  of the measurement box outside the scanned region...these are not
  actually integrated unless the PROFILE EDGE keywords are included, in
  which case those with at least half of the pixels inside the scanned
  area will be measured by profile fitting). Overloaded reflections are
  also not integrated (by default) but if the PROFILE OVERLOADS keywords
  are included, then their intensity will be estimated by profile
  fitting the non-overloaded pixels. NNEG is the number of reflections
  with negative (summation integration) intensity. The proportion of the
  dataset that falls into this category obviously depends on the
  strength of the images.

  BAD

  The number of badspots. Reflections will be classified as badspots if
  they fail any of five criteria: BGRATIO greater than 3, PKRATIO
  greater than 3.5 (Fully recorded reflections only, and the profile
  fitted measurement only is flagged), intensity negative and greater
  (absolute value) than 5 sigma, background gradient too large 
  (background gradient/ average background exceeds 0.03),
  too few background pixels remaining after
  rejecting outliers (less than 10).  All these criteria (except
  negative intensity) can be changed using the REJECTION keyword.  There
  should be less than 5-10 badspots on any one image. If there are more
  it suggests problems. The badspots are listed in the logfile with the
  information necessary to see why they have been flagged. It may also
  be helpful to have the pixel values of flagged spots. These can be
  obtained by using the keywords: REJECTION PLOT.  This should
  enable the source of the problem to be identified.

  I/SIGI (two columns)
  This gives the average I/sd(I) for the whole dataset (first column)
  and the outermost resolution bin (second column).





1 Example
     An example command file is given below. See the MOSFLM User Guide
  for other examples.
P
  ipxmosflm  spotod  /scr0/andrew/f1.spotod \
                        summary f1x100_1to30.sum \
                         << eof-mosflm_ip
  TITLE F1 native data station 9.5 SRS 14/9/93
  HKLOUT f1x100_1to30.mtz
  GENF  f1x100_1to30.gen
  IDENT f1x100
  MATRIX f1x100_001.mat
  MOSAIC 0.1
  SYMMETRY P212121
  SYNCHROTRON
  DIVE 0.101 0.007 
  DISPER 0.00020 0.0
  DISTANCE 330
  WAVELENGTH 0.92
  SEPARATION 0.8 0.8
  RASTER 13 13 6 4 4
  PROCESS 1 30 ANGLE 0.5 START 0.0 ADD 1000
  GAIN 1.2
  ADCOFFSET 8
  BEAM 90.9 91.2
  EXTENSION pck
  DIRE /hx0/andrew/
  DISTORTION ROFF -0.44 TOFF -0.40
  RUN
  eof-mosflm_ip
p