The HIRES spectroscopic efficiency has been measured with both the UV and red-blazed cross dispersers, by acquiring "slitless" spectra of spectrophotometric standard stars. These measurements were made on the nights of September 25 and November 11, 2005 (UT) respectively. Conditions appeared to be photometric on both nights. The entrance aperture was a custom 4 arcsec wide and 5 arcsec high decker hole. Comparison spectra were taken for all wavelength settings using a normal 0.57 arcsec wide slit and used to determine the wavelength scale.
On September 25 (UT), with the UV-blazed cross disperser (HIRESb), two wavelength settings (XD angles 0.90 and 2.08) were used to tile the region from 3000 to 6700 Angstroms. HIRESb observers are likely to use the image rotator to mitigate the effects of differential atmospheric refraction. We used the rotator set to the sky parallactic angle on this night. Four, 300-second integrations were obtained at each wavelength setting of BD+28d4211. Three, 300-second integrations were obtained of Feige 110 at only the longest wavelength setting as a consistency check, and because BD+28d4211 has a very faint red companion. The Feige 110 and BD+28d4211 data sets agreed almost exactly. No order sorting filter was used. At the very longest wavelengths measured, the second order is faintly visible, but affects the measurement at a level less than one percent.
On November 11 (UT), with the red-blazed cross disperser (HIRESr), two wavelength settings were used to tile the region from 4100 Angstroms to 1.02 microns (XD angles 0.43 and 1.67). HIRESr observers may or may not use the image rotator depending on their preference since at longer wavelengths, the effects of differential atmospheric refraction are not as great. We did not use the image rotator on this night. Order sorting filters were used since this is almost always required at wavelength settings chosen by those using HIRESr. Eight, 300-second integrations of G191B2B were obtained at the longest wavelength setting. At the shorter wavelength setting, several sets of three, 300-second integrations were obtained using a number of different order sorting filters to allow measurements on the different CCD's of the detector mosaic.
All the data described above were measured and analysed the same way. Each order was measured at the peak of the blaze function. The total number of counts within 25 pixel square boxes was measured using the IRAF imexam task. The background, which includes bias, dark current, scattered light, etc. was subtracted. Knowing the integration times, gain factors, and dispersion coefficients allowed a conversion to detected photons per second per Angstrom. These numbers were divided by the (primary*secondary*tertiary) telescope reflectivity. Using the spectrophotometry and formulae from Massey et al (1988) (ApJ 328, 315) combined with tabulated extinction tables for Mauna Kea, and knowledge of the collecting area of the Keck I primary allowed the calculation of expected number of detected photons and finally the efficiency.
In the plot below, measurements made on the different CCD's of the mosaic are indicated by the different colour symbols.
Plot of HIRES Spectroscopic Efficiency