It is not always easy to identify ``the'' time of an observation. Some instruments record the time at which the observation began; some record the end of the measurement, or even the time at which the readout was recorded; very few record the middle of the exposure. The ESO Archive records the starting time of the observation, recorded in a FITS-header keyword named MJD-OBS. Because of problems in MIDAS with strings containing embedded minus signs, the column label used here is MJD_OBS. This quantity is the geocentric Julian Date at the start of the integration, minus 2400000.5 days. To retain adequate precision (1 second is required), this must be stored in double precision.
Note that this is geocentric, not heliocentric MJD. It is not suitable for use in computing phases of eclipsing binaries and the like without light-time corrections.
Having accepted a starting time as the basic timing datum, we must in every case have an integration time, even for data (like DC photometry) where this is not directly involved in calculating signal strength. In any case we would need this information for weighting purposes. Again the ESO Archive name is used as column label: EXPTIME.