2 Strategy of observation

The studied region is $\sim 20 \times 24$ arcmin wide, located $\sim15$ arcmin North-East from the Coma cluster center (see Fig. 1). The approximate center of the region is 13 00 35 +28 14 35 (J2000). We voluntarily avoided imaging the Coma cluster core, because of the possible presence of a halo around the two dominant galaxies, which would make the reduction and analysis of the data more difficult, given the small field of view of our camera.

Data were taken in the near-infrared H band, on March 4th and 9th, and April 3rd to 5th, 1997, at the 2-meter Télescope Bernard Lyot of Pic du Midi Observatory (France), with the Moïcam camera set at its f/8 focus. Moïcam is a $256\times256$ pixel camera, based on an array of $2 \times 2$ NICMOS-3 detectors. The adopted configuration gives the maximum field of view available, $2'~\times~2'$, with a pixel size of 0.5 arcsec. A few hours per night were allocated to us in March, and half-nights in April.

For an efficient use of the assigned observing time, we did not guide the telescope during the exposures. This limits the individual exposure times to 30 s, without appreciable degradation of the PSF. This is also an optimum value for the exposure time in H, in order to reach a suitable signal-to-noise ratio on individual exposures, while keeping the sky background relatively low ($\sim3000$ to 6000 ADU). In order to image the $20\times 24$ arcmin region with the camera, we mosaicked the total field by successive strips in declination, moving the telescope by half a field of view in declination between exposures, and by half a field of view in right ascension between strips. In this way, with the exception of the regions close to the edge of each field, each point in the sky is observed four times, giving a total exposure time per sky pixel of 120 s.

Almost half of the field was observed more than four times in order to reach fainter magnitudes. This was achieved in three ways: by doubling the number of exposures at each telescope pointing, by moving the telescope by 1/3 of a field of view instead of 1/2, or by observing the same strip on different nights. The final median exposure times are 310, 120 and 400 s respectively for the fields patch1 (0403+0903+0304), 0404 and 0504, as were dubbed the regions - from their date of observation.

Unfortunately, the telescope does not point with a precision of one pixel, and therefore, our 675 manually introduced telescope offsets are only approximately achieved on the sky (see Fig. 2). These errors accumulate to create a field which is not exposed in a perfectly uniform way, and, since the actual offset is a few arcsec smaller than the required one and systematic errors accumulate in RA, there are regions exposed more than required and others never exposed. This problem was unknown at the time of the observations, probably because no-one had tried to use our observational strategy before at this telescope (we observed during the very first observing runs with the NICMOS-3 camera). However, since we returned repeatedly to the same regions, only one narrow gap is left in the whole field, between fields 0504 and 0404. Two bright galaxies of Coma fall in this gap, GMP 2413 and GMP 2418, as well as several QSOs and stars.

All images have been taken with the same exposure time, 30 s, in order to simplify the data reduction.

Figure 4: Average central surface brightness vs. magnitude in a 10 $^{\prime \prime }$ aperture for the three fields. The completeness magnitude is indicated by the horizontal dashed line. Circles mark reliable galaxies, crosses objects with uncertain star/galaxy classification (0.7< stellarity index <0.8)