s, another type of Cerenkov counters
will be installed behind the second TOF walls.
Figure 26 shows the refraction index dependence of the
number of photo electrons for proton and K from the Cerenkov counter.
In our momentum region of interest (around 1.2 GeV/
), a proton emits
over 50% Cerenkov photons of those for K in lucite (n=1.48); hence,
the on-line separation of proton was not easy with the lucite Cerenkov counters
for the E89-009 experiment. Figure 27 shows estimated ADC spectra
for water and lucite Cerenkov counters with a Monte Carlo simulation.
Cerenkov light emission of proton in water is much smaller because of
the smaller reflective index of 1.33 and, thus, water Cerenkov is better
for the K, p separation in this momentum region.
Two layers of counters (HWC1 and HWC2) are necessary to achieve
on-line rejection efficiency of
for protons,
which corresponds to the proton contamination rate in the kaon trigger below 100 Hz.
Wavelength shifters such as amino-G-salt and 4-methyl-umbelliferone could be doped in water
to make light response isotropic and to increase number of Cerenkov photons
unless they deteriorate proton rejection efficiency.
Simulation study and test with a prototype are in progress.
A test experiment for the prototypes of the
aerogel, water and lucite Cerenkov counters is scheduled at KEK-PS with 1
2 GeV/ p and 
in winter, 2001 (KEK PS-T494).
With the result of this test beam time, the design of Cerenkov counters
and option of the radiator (water or lucite, with and without wavelength shifter) will be finalized.