In general, the galaxy luminosity function (LF) 
is
approximated by the analytic Schechter (1976) form:
where 
, 
, and 
are the normalisation of the
number of galaxies per unit volume, the index of the faint-end slope,
and the characteristic luminosity, respectively.
Of particular interest to the LF are elliptical and lenticular (E/S0)
galaxies, which dominate the contribution to lensing statistics (Maoz
& Rix 1993). Many models of gravitational statistics use the
Schechter-form LF for all types of galaxies from Efstathiou et
al. (1988, hereafter EEP) scaled by the fraction ( 
30%) of E/S0 galaxies
reported by Postman & Geller (1984). CY note, however, that due to
the different sampling and calibration of the two studies, unknown
systematics in their combination likely produce
uncertainty in the details of the E/S0 LF shape. To correct this, CY
consider the Schechter parameters for LFs determined using only E/S0
galaxies in the Stromlo-APM survey (Loveday et al. 1992, hereafter
LPEM) and the CfA survey (Marzke et al. 1994, hereafter MGHC).
The lensing probability can then be computed as a function of the LF
parameters. In addition to depending on the faint-end slope parameter
,
the probability of lensing is proportional to the fourth power of the
velocity dispersion 
, which can be
determined from the characteristic luminosity using the
Faber-Jackson relations (de Vaucouleurs & Olson 1982):
where 
is expressed in magnitudes, and 
for ellipticals and 19.75 for lenticulars.
The lensing probabilities thus determined are lower for both the LPEM and MGHC LFs than for the EEP LF. By contrast, K96 mixes the parameters from the scaled all-type EEP, LPEM, and MGHC LFs and obtains a higher lensing probability than that for the EEP LF. In addition to their cautionary note about scaling all-type LFs, CY note that Schechter parameters for a given survey are determined in a highly correlated manner and so mixing the results of several surveys in a contrived fashion is inconsistent and introduces artificial systematics.