Pump-probe spectroscopy is a common method in the study of ultrafast relaxation processes. In the study of the relaxation of hot electrons in semiconductors, there are two possible approaches using this method, each of whici yields different data about the relaxation processes. The first approach probes the relaxation of a population of hot electrons, which has previously been created by photoexcitation, by measuring the saturated absorption of a probe pulse which has a longer or shorter wavelength than the exciting pulse. As the wavelength of the probe is tuned, a picture of how the electrons move down the central valley is obtained.1,2 Some of the electrons, however, may not remain in the central valley as they lose energy but could be scattered to the satellite valleys, at least for short times. Since carriers in the satellite valleys are not connected by a direct optical transition to the valence band at typical photon energies, even an experiment which uses a tunable probe wavelength will not be able to probe all the possible states into which the electrons may scatter. The second approach is complementary to the first. Rather than probe the passage of electrons through states of lower energy than the initial state, the way in which they leave the initial states is determined by measuring the saturated absorption of a probe pulse of the same wavelength as the exciting pulse. A complete picture of the electron relaxation requires the results of both approaches. We have chosen to pursue the second scheme. A point that is often overlooked is that rough numbers for the scattering rates involved already exist, and if time-resolved spectroscopy is to add anything new it must be quantitative. This means, of course, that the data must have a high signal-to-noise ratio. One advantage of the second approach is that the high repetition rate of the pump laser may be put to good use in reducing noise in the data. Because the pump and probe pulses have the same wavelength, they may be generated directly from the output of the laser, and consequently the repetition rate of the experiment is equal to that of the repetition rate of the laser pulses, or about 100MHz. This leads to an increased signal-to-noise ratio as compared with the tunable pump-probe method, in which the pump and probe repetition rates are generally about 10 kHz.