An evaluation has been made of some of the errors introduced into the electron diffraction of gases by failure of the sample to be localized, and by secondary and higher order scattering.The degree to which poor nozzle design may contribute to these factors was investigated experimentally by measurements of the pressure in the camera and absorption measurements of the main electron beam when a small volume of gas was allowed to expand into the camera; and theoretically for the cumulative effects of higher order scattering on the final picture. This evidence was complemented by density measurements on photographic plates taken under similar conditions. The effect of using liquid air cooled surfaces around the nozzle was tested.Of the nozzles tested, better ``image to background'' ratios were obtained with the ``hydrodynamic'' and ``comparison'' nozzles than with the ``conventional'' nozzle, but the gas must be admitted slowly enough so that less than about 10 percent of the main beam is appreciably scattered. The ``image to background'' ratios were always higher using adequate cooling on the condensing surfaces surrounding the nozzle. In the case of pressure and electron beam absorption measurements short exposures were advantageous, where secondary scattering was not directly considered, and high camera pumping speeds were always advantageous.On the basis of these results some general design suggestions for electron diffraction cameras, and requirements for good efficiency in nozzles can be made. A nozzle design is proposed which would reduce materially the errors normally encountered in electron diffraction work.