3 Observations

Imaging observations of the clusters were carried out during the period from May of 1992 through January of 1994 using the University of Hawaii (UH) 2.2 m telescope equipped with a Tektronix 2048 $\times$ 2048 pixel thinned, back-illuminated, anti-reflection-coated CCD mounted at the f/10 Cassegrain focus. The image scale was $0\hbox{$.\!\!^{\prime\prime}$}22$/24 $\mu$m-pixel and the field of view was $7\hbox{$.\!\!^{\prime\prime}$}5 \times 7\hbox{$.\!\!^{\prime\prime}$}5$ (1.54 h₅₀^-1Mpc $\times$ 1.54 h₅₀^-1Mpc at z=0.15 and 3.74 h₅₀^-1Mpc $\times$ 3.74 h₅₀^-1Mpc at z=0.82). R band images of all, and B band images of most of the clusters were acquired using this configuration. We also took additional V and I images of a few selected clusters using both the UH 2.2m and HRCam (McClure et al. 1989) on the CFHT. In total, there were 8 observing runs: May, June, September and October of 1992, February, May and November of 1993, and January of 1994. The majority of the data are of excellent quality. Except for the February 1993 run, the images were taken in photometric conditions and in good seeing. The R seeing ranged from $0\hbox{$.\!\!^{\prime\prime}$}5$ to $1\hbox{$.\!\!^{\prime\prime}$}3$ FWHM with a median value of $0\hbox{$.\!\!^{\prime\prime}$}8$ FWHM. The seeing for the B images was slightly worse with a median of $0\hbox{$.\!\!^{\prime\prime}$}9$ FWHM.

We chose the B and R bandpasses for the following reasons. The R bandpass is ideal for observing the distant clusters in the redshift range spanned by the EMSS sample. Furthermore, the center wavelength of the R filter occurs near the peak of the CCD quantum efficiency, and the night sky is darker than in a redder bandpass such as I. Although gravitationally lensed arcs are known to be relatively blue, they are still easily detected in the R band. The additional B or V band images were taken to allow us to recognize any gravitational arcs by their relative blue color compared to the red cluster galaxies.

In order to build up our long exposures, we used the standard "shift-and-stare'' or "dithering'' technique where we took a number of short integrations (typically 600s or 900s each) with the telescope shifted $\sim$20-30'' between exposures, allowing us to assemble a median-filtered stack of the disregistered images to use for flattening the data. Typically, the set of all images in a single color spanning an entire night, and often an entire run ($\sim$50-100 frames), were used to build the median skyflat. Each separate cluster image was first bias subtracted and then flattened with the normalized median flat. The final image was then produced by shifting the individual images into registration (integer pixel shifts) and adding them while cleaning them of cosmic ray hits.