every case the average values are in good agreement with the correct values. This indicates that there is no appreciable systematic error in the method, and reliable absolute concentrations may be determined if replicate runs are made.More work is needed to uncover the source of scatter in the data. Volumetric and gravimetric errors did not exceed 2% each. The isotopic ration measurements are accurate to about the same figure, as shown by measurement of both the singly and doubly charged ion ratios. Loss during sample dissolution does not appear to be a consistent explanation because it would produce systematic errors, not just scatter. One possibility is that chemical equilibrium between the components of the spike solution and the corresponding components of the sample solution does not occur before addition of the graphite and subsequent heating to dryness. This could result in selective distillation during the drying process of one component or the other, depending on conditions. Another possibility is immediate precipitation upon addition of the spikes.The technique investigated here is the simplest and fastest possible. If several samples are prepared a t one time, total preparation time per 10 element sample would be about 1% hours. Only about 5 minutes is required to cut a sample for direct sparking. However, no standard is required for the isotope dilution method, so that the overall analysis time is about the same as for direct sparking with electrical detection.Assuming that the precision can be improved, the main disadvantages of the isotope dilution method are: 1) Two of the most important elements in bullet lead, As and Bi, cannot be determined. 2) An order of magnitude estimate of the elemental concentration is needed in order to spike a t the optimum level. This means that the isotope dilution method could best serve as an adjunct to the direct sparking method, yielding absolute concentrations for those elements for which no standards are available and providing a valuable check on other elements.In the comparison of a 42 cm3 Ge(Li) and a standard 7.6 X 7.6 cm Nal(TI) detector for the y-ray spectrometry of low intensity, non-complex sources, it has been shown that there exists some source intensity, lcritlcai, for which two compared detection systems yield identical precision for identical count time. lcritlcai was found here to be -1 photon sec-' between 140 and 1600 keV. Below this intensity, the Ge( Ll) system always yields superior precislon whereas Present address, Department of Chemistry, McMaster University, Hamilton, Ontario, Canada. above lcrnlcal the Nal(TI) system Is ascendant. This result is contrary to that expected using arguments based on counting efficiency considerations alone. Optimum peak integration limits have been derived for low signal-to-noise ratio conditions assuming a Gaussian peak shape. Even in the most unfavorable case, where the signal-to-nolse ratio is approaching zero, integration should be performed across at least 84% of the peak area, Le., between limits of f1.4...