The present investigation sponsored by the Illinois Institute of Technology, Armour Research Foundation, and Institute of Gas Technology, deals with the heat transfer of air in parallel flow to a surface, a process of great practical importance which, for instance, occurs with all kinds of fins or on the skin of an airplane in flight. Previous knowledge of that process was based on a few sets of experiments which were performed with plane surfaces and led to a considerably higher heat transfer than a reliable theory, due to H. Latzko. In particular, the influence of nonheated starting sections seemed to require a new investigation. Compared with the use of plane plates by previous investigators, the use of an electrically heated cylindrical specimen has the following advantages: A cylinder can be easily placed in the center of an air jet and is free of the edge losses of a plate; for both reasons, air jets of moderate diameter can be used. Uniform heating is easier to provide, heat losses to the back are easier to control, and noses of different shape and cylindrical starting sections can readily be used for studying the behavior of the so-called boundary layer of the fluid which is developing along the surface, first streamlined and then turbulent, and in which all resistance against heat transfer is concentrated. The experiments were performed with specimens of 1.3 in. diam and 9 to 20 in. total length, the ratio of the heated length to the total length being varied from 40 to 90 per cent. Spherical, ellipsoidal, and conical nosepieces were used. The air velocity was varied from 10 to 150 fps and it is assumed that the results can be theoretically extrapolated above sound velocity, using the principle of similarity.