1) Documentation (RO_MIR_IF_0001.PDF) says that software is supposed to
   be in the SOFTWARE directory, but the only supplied software is
   actually in the DOCUMENT directory.  There is the following (in that
   PDF file):

   3.4.3.9 Software Directory
   It is intended that the software used to calibrate the data will be
   included in this directory for level-3 products. Currently, this
   software is in an early stage of development and is tied to the
   local database used by the processing, hence is not suitable for
   delivery to the archive at this time. The description of the
   algorithms in section 2.3.3.2 fulfills this function, for now.
   Therefore, this directory was omitted in the current deliveries.

   but, that software *is* delivered, in the DOCUMENT directory.

2) That software was very hard to run sensibly; it didn't supply
   meaningful defaults for hardly any of the inputs, seemingly one has
   to know quite a bit about the data to make it return what you want.
   For instance, it asks you to "enter 1 if UTC field is present, 0
   otherwise:".  I have no idea whether the UTC field is present or
   not in each of the datasets, or what it means if it is or isn't.
   In addition, more comments in the code would be extremely useful.
   on the other hand, this software *was* extremely useful - it was
   a quick way to gain access to the data.  During the steins review
   Amy had tried to use the readpds function of IDL to read this
   data and had failed - I'm not familiar with that so couldn't test
   it, but for me this FORTRAN program was great.

3) There is a column in the continuum data that is supposed to indicate 
   that a calibration is occurring.  From the format file for the
   continuum data:

   OBJECT       = COLUMN
     NAME             = CALMODE
     COLUMN_NUMBER    = 12
     DATA_TYPE        = MSB_UNSIGNED_INTEGER
     FORMAT           = I1
     START_BYTE       = 60
     BYTES            = 2
     DESCRIPTION      = "0: Calibration in progress, 1: No calibration in
   progress"
   END_OBJECT   = COLUMN

   but in the actual data, there are many entries in that column that
   are 256, not 0 or 1.  What does 256 mean?

4) There are some discontinuities in the mm data - what are they?
   (see plots.)  Similarly, very odd things happening with the submm
   data in the early pre-encounter period (and similar discontinuities).
   what's up with that?

5) It looks like the encounter is contained in one file only, but there
   are two for each receiver.  For instance, for the mm Rx, 
   MIRO_3_MM_20101910614.DAT contains the encounter.  What does
   MIRO_3_MM_20101920345.DAT contain and why do we care (it's clear
   from the timestamp in the filename that it's from a later time,
   but not clear what it is for)?  Might be in the documentation 
   somewhere - haven't found it yet.

6) In the spectral data there are large discontinuities between the
   seven "bands" (see plots).  I would have expected this to have 
   been removed in level 3 data?  there are also many weirdnesses in 
   the spectral data (see movie.)

7) The method for calculating brightness temperature, the mapping from
   spectral channel number to sky frequency, and details on geometry
   are all either clearly outlined in the user manual or contained in
   the data release itself.  This is refreshing given their lack at the 
   time of the Steins review.