15   -3
NA > 5x10   cm- . While in general, agreement between theory and experiment in this region of dopant density is only within 15 percent, the bulk of the experimental data is within 10 percent of the theoretical prediction. The samples used by Morin and Maita [5] showed a considerable degree of compensation; in their lower-doped samples, the compensation was as high as 20 percent. This fact may have contributed to the low values of measured Hall mobility, and the large discrepancy between these data points and the calculation. This model neglects the effects of compensation, and the combined presence of acceptor impurities of varying ionization energies. The overall effect of adding impurities of both signs is one of increased ionization of the excess NA or ND [15]. It would be necessary to know the percentages of compensation of the experimentally measured crystals to accurately determine the adequacy of the theory at low dopant densities. An experimental estimate of percentage of compensation was.not made for any of the silicon samples studied in this work. Long [38] has noted that the low measurements of Hall mobility for p-type silicon may not be due entirely to compensation and the quality of the crystals. For reasonably pure silicon samples (p = 35 ohm-cm, NA = 4.4xi1 cm-) Long [38] obtained Hall mobilities between 360 and 390 cm2/volt-sec. Hall mobilities as great as 450 cm 2/volt-sec have also been reported [76]. While higher than the Hall mobilities of Morin and Maita, these measurements [38,76] still indicate a value of Hall factor for low-doped p-type silicon at 300 K less than unity. It is still doubtful that a Hall mobility smaller than a conductivity mobility at 300 K is really an intrinsic property of p-type silicon [38]. It is possible that the Hall factor may be greater than unity in a crystal of exceptionally high perfection. However, calcu-