spacing automatically, but the method outlined above is considered inherently more accurate. The leading edge of the neutron pulse could also be transcribed to film. Here, the transient response of the canister system is very {mportant, and studies of this are not complete. However, it should be possible to determine a reasonable equivalent transfer function for use in obtaining the detector-current function giving rise to this output pulse. This procedure will give only the shape of the leading edge of the pulse. The amplitude must be determined from the pulse rate at the time the system reverts to converter re- cording. This difficulty arises from the inability of tape recorders to faithfully reproduce amplitude recorded information. Consecutive recorded pulses may vary as much as a factor of four in amplitude, although the average amplitude ahould remain accurate to about § percent. The accuracy possible with the system should be of the order of + 25 percent for the converter information at low frequencies, increasing with frequency. It was expected that neutron-pulse-rise information would have been worse than this and would not have been capable of evaluation until the transfer function investigation had been completed. 33