Furusawa-san and myself looked at PFSA06928311.fits and PFSA06928312.fits and found the following issues and concerns. I think that some may not be urgent for the November 2021 run and some require Subaru FITS committee to work on the observatory policy, but let me list all of them here.

1. Basic keywords in the primary HDU

1. DISPRSR should be DISPERSR
2. DARKTIME is not in the dictionary (but maybe there is a Gen2 alias?)
3. OBS-ALOC is defined to be either Observation or Standby (need to check if the Gen2 communication is valid or not)
4. RA_CMD is not in the dictionary (but there is a Gen2 alias)
5. DEC_CMD is not in the dictionary (but there is a Gen2 alias)
6. ADC-TYPE is defined to be either NONE or BLUE

• Those not in the Subaru's dictionary should be registered by the Subaru FITS committee, if there are Gen2 aliases and they are necessary to be in the header.

• The same is true for OBS-ALOC and ADC-TYPE. If the current values are more appropriate, please let us know.

Format issues

DEC, DEC2000 A12
GAIN F20.3
DET-TMP F20.2
WAVELEN, WAV-{MIN,MAX} F20.4
SEEING F20.2
TRANSP F20.3
LONPOLE F20.1
EQUINOX F20.3
AZIMUTH F20.5
ALTITUDE F20.5
ZD F20.5
AIRMASS F20.3
INST-PA F20.3
INSROT F20.3
FOC-VAL F20.3
ADC-STR F20.3
DOM-PRS, OUT-PRS F20.2
DOM-WND, OUT-WND F20.2
M2-POS[1,2,3], M2-ANG[1,2,3] F20.3

• It is fine to be shorter than the specified formats. For example AIRMASS=1.52 is fine for the "F20.3" format, but it would be nice to check if they are designed not to exceed the specified length.
• Formats in the dictionary can also be revised if the precision higher than that defined in the current rule is necessary.

2. Instrument keywords in the primary HDU

1. Unit [C] may be replaced by [degC]: W_CRTEMP, W_CRTEMP, W_CRSETP, W_XCL2RJ, W_ETMP1 , W_ETMP2 , W_ETMP3 , W_ETMP4 , W_ETMP5 , W_ETMP6 , W_ETMP7 , W_ETMP8 , W_ETMP9 , W_ETMP10, W_ETMP11, W_ETMP12, W_ETMP13, W_ETMP14, W_ETMP15, W_ETMP16, W_ETMP17, W_ETMP18, W_ETMP19, W_ETMP20
2. Temperature entries for W_XCL??? use "[degC]" and "[K]". Are they as intended?
3. Is the datatype be boolean?: W_XH1ENA, W_XHP1EN, W_XH2ENA, W_XHP2EN
4. "0..1" in the comments should be "0 or 1"?: W_XH1FRA, W_XH2FRA
5. No comment is assigned: W_ENFCAS, W_ENFCAT, W_ENBSHS, W_ENRDAS, W_ENRDAT, W_ENTMPS, W_ENTMPT

• Datatypes and formats are not fully checked as I don't have the up-to-date dictionary.

3. Queue-related keywords

There are keywords for queue observations (https://www.naoj.org/Observing/fits/header/queue_keywords/). Not all of them are valid for PFS, but need to consider to put some of them. For example, OBS-MTHD, MOON-ILL, MOON-SEP, and MOON-EL will be helpful to distinguish the observing mode, but PROP-PI would not be appropriate as fibers are typically shared by multiple PIs.

4. Basic keywords in the first HDU

1. Is the location of INHERIT correct? The FITS standard says that if present it shall appear in the extension header immediately after the mandatory keywords. We need to check which keywords are actually mandatory, but if you have any thoughts, please let us know.
2. If I remember correctly, putting WCS as pixel coordinates are agreed or recommended by us after a long discussion, though the FITS committee think that it is still useful to put even a rough wavelength solution with CTYPE2=WAVELENGTH.
3. Do we need BLANK=-32768 and how it can be set in the case of tiled-image compression?
4. CUNIT1 and CUNIT2 should be inserted as "pixel" or "pix"?

5. Missing spectroscopic keywords required by Subaru FITS regulation

1. There are several keywords related to slit properties. As PFS is a fiber multiobject spectrograph, it's not clear what is the appropriate entries. Subaru FITS committee also does not have a clear regulation on it. As a reference, corresponding keywords are used as follows in FMOS. I think SLT-LEN and SLT-WID look useful also for PFS (either in [arcsec] or [mm]?)

SLIT    = 'FIBRE-MOS'          / Identifier of the entrance slit used

SLT-LEN =                  1.2 / Diameter of the fibre entrance (arcsec)

SLT-WID =                  1.2 / Diameter of the fibre entrance (arcsec)

SLT-PA  =                   90 / Field position angle (degree)

SLTCPIX1=                 1024 / Slit center projected on detector (pixel)

SLTCPIX2=                 1024 / Slit center projected on detector (pixel)