The precision of measurements on a photographic plate depends on the image area. But the dimensions of the images given by a grating spectrograph, a Fabry-Perot spectrograph and the SIMAC are very different. So it was interesting to develop a new comparison which takes into account not only classical illumination but also image size. Then, we shall first discuss, at a given resolution the optical arrangement which gives the greatest image area. Secondly, we shall evaluate the signal upon noise ratio dependence with the area and the optical density of the recording. - Greatest image obtention Grating spectrograph: The best illumination is obtained when, along the direction of dispersion, the image width is equal to the graininess « g » of the photographic emulsion. Its heigth is not definite, but, because of the curvature of the image, experimentalist use small part of the entrance slit. We have choosen following limitative criteria: the angular heigth must be such as the sagitta is equal to the limit of resolution. (This definition gives values in good agreement with practical ones.) Classical Fabry-Perot spectrograph: The width of rings must be equal to the « g » value: this condition determines the focal length of the objective located after the Fabry-Perot. Focal length of the grating spectrograph which assumes the separation of orders must be sufficient to assume the R/N resolution. (N coefficient of « finesse »). In fact we demonstrated that it was better to increase this length so that «etendue» (or acceptance) of the Fabry-Perot and the of the grating are adapted to each other. If optical magnification of the grating spectrograph is fixed to unity, slit must be enhanced. With this process, illumination and resolution are inchanged, but the useful part of the ring is more important. SIMAC: Because of the most important « etendue » delivered by the Fabry-Perot interferometer, it is possible to adapt a short focus grating spectrograph. But then, the total Fabry-Perot has to be illuminated to get the hightest possible image. Under these conditions, calculatins shows that, at a given resolutin, these three spectrographs deliver quite the same energy by unit time. - Signal to noise Ratio Clark Jone's and Felgett's works etablish the existence of a meximum of the detective quantum efficiency for the photographic receiver. For a given emulsion, there is an optimal density D0. So, we can separate the discussion into two parts. In the first case, we assume that we can reach in a reasonable time the optical density D0 with the slowest instrument. Under these conditions, if we use the Selwyn's law (density fluctuations are inversely proportionnal to the square root surface) and the reciprocity law, we see that the signal to noise ratio is varying as the square root time, exactly as in spectrometry. So, in this case, the three instrument are quite equivalent. The gain in rapidity is strictly compensated by a loss in signal to noise ratio. If we increase the image area of the Fabry-Perot spectrograph or of the SIMAC, we shall have the same exposure time and signal to noise ratio as with the grating spectrograph. Note, in this purpose that in the case of the SIMAC the change of focus length is not necessary because of the semi-sequential recording. In the second case (D0 is not accessible in a reasonable exposure time), the study of the variation of signal to noise ratio against density demonstrates that it is better to choose a short focus instrument. The decrease of the signal to noise ratio is compensated by the gain of detective quantum efficiency especially for very weak illuminations.