Measurements of spectrometer resolution using an electron gun and phosphor type of light source yield values considerably better than those obtained from scintillations in a phosphor. Line widths of about 3.4 per cent have been obtained at a pulse height equivalent to 661 kev in NaI. An investigation is being made to determine the causes of this difference. A number of possibilities have been eliminated. Flasher light pulses may be collimated or diffused at the photo surface with no difference in result. They may be mixed with random scintillation pulses giving simultaneously the relatively poor resolution of the scintillator and the good resolution of the flasher. Optical filters do not affect the result when the flasher is used, but do give a relative improvement with a scintillator. It is found that as the light is attenuated in any way from a scintillation crystal, its relative resolution is improved. The improvement is compatible with the concept of an intrinsic crystal resolution. In other words, the difference of the squares of crystal width and flasher width remains constant with varying attenuation. It remains constant also for a given crystal with photomultipliers of different types and resolutions. The intrinsic width indicated for a good crystal is about 6 per cent. Results of experiments attempting to locate the cause of this effect are reported.