The lower limit of size of biological objects which can be photographed with the electron microscope, by the aid of shadow‐casting, is shown to depend upon the smoothness of substrate upon which they can be mounted, and upon the continuity of structure of the thin films with which they are shadowed. Numerous attempts to improve the existing deficiencies are reported, both with respect to producing smoother substrate films, and to producing films for shadow‐casting of high efficiency and continuity of structure.No success has been encountered in producing a usable substrate film perceptibly smoother than the collodion and Formvar films commonly used. It is found that the best shadow‐casting material for this type of film is uranium or uranium oxide.Verification has been obtained of the severe granulation of gold films previously used in the pre‐shadowed replica process, when subjected to the electron current of a biased‐beam electron gun. Attempts to reduce the granulation to a satisfactory level have failed. Various methods of preparing pre‐shadowed replicas are reported. The factors affecting the tenacity of evaporated films to glass surfaces are discussed. It is found that elements which oxidize readily are relatively adherent to glass, while the elements gold, palladium, platinum, and rhodium are not. Uranium sulfide can be used as a pre‐shadowed replica material, but only with some uncertainty, owing to its chemical instability. It has been found that a palladium‐platinum mixture is the most satisfactory material for use in the pre‐shadowed replica technique, and that films of this mixture in a thickness of about 6A produce adequate shadows in which there is no sign of granulation. The surface of clean glass is again found to have the smoothest structure of any material known, with practically no sharp discontinuities in elevation as great as 10A.An appendix is given, in which technical details of shadow‐casting and replica production are described.