AS ANGIOGRAPHIC diagnosis becomes more and more refined there is a demand for radiographic demonstration of smaller and smaller vessels. The various factors restricting resolution of x-ray images are well known, and it is proposed here to make some comments on the problem of geometric unsharpness and the use of primary or direct magnification radiography in vivo. Optimal resolution of vascular detail is obtained by low kilovoltage radiography of thin specimens in contact with fine-grain emulsion film, followed by photographic enlargement. In order to apply secondary or photographic magnification to in vivo angiography, the organ to be studied must be relatively thin and placed in direct contact with the film-holder. Margulis used this method in small experimental animals and on the human bowel at surgery (5). Recently Bosniak et al. applied this method to the surgically exposed kidney and bowel of dogs, using industrial-type film (3). In both studies, film grain limited enlargement to tenfold at most. Because of limitations on the rapidity of exposure and excessive radiation dosage with fine-grain film, as well as the inconvenience of the photographic step, secondary magnification has few clinical applications. Primary or direct magnification radiography does not require fine-grain film or special screens so that the overall quantity of radiation need not be excessive. The limiting factor is geometrical blurring due to the size of the focal spot relative to the object studied. Focal spot size is limited on the small side by the concentration of heat on the anode during the x-ray exposure. Experimental tubes with ultrafine focal spots have been described but find little clinical application because of the severely restricted heat capacity (9). Recent interest in direct magnification has been stimulated by the development of x-ray tubes with improved heat dissipation and relatively small focal spots, made possible by such features as decreased anode surface angulation and increased rotor speed and surface emissivity. As a result, studies can now be performed on the human body, through even its thick parts, with acceptably short exposures, at two- to fourfold magnification (4, 8, 10). On the basis of simple geometric considerations' it is possible to estimate the size of the smallest object, relative to the focal spot, that could be demonstrated with a given primary magnification, assuming that only the umbra image is useful. Bellman noted that an image composed only of penumbra would have poor contrast (1). Milne suggested that the apparent size of an object smaller than the focal spot would actually diminish with increasing magnification, referring, of course, to the umbra component only (6). Takahashi described the disappearance of objects smaller than the focal spot as a “filter effect” (9). The validity of these theories depends on whether an image could indeed be usefully detected if it were made up entirely of penumbra.