This atlas contains intensities, expressed in terms of the continuum as 100, in steps of 5 milli Angstroem of the F5 star Procyon. The spectral range is from 3140 to 7469.995 Angstroems. Introduction: The Atlas covers the spectrum of Procyon from 314.0 nm to 747.0 nm. Spectra from 314.0 nm to 540.0 nm are generally exposed on fourth- and third-order plates and from 540.0 nm to 747.0 nm mostly on second-order plates. All wavelengths are in the rest-frame of the star, i.e., they have been reduced to zero radial velocity. This is one of three classical spectrophotometric atlasses, the others being, 'A Photometric Atlas of the Spectrum of Arcturus' (Griffin 1968), and the Utrecht 'Photometric Atlas of the Solar Spectrum' (Minnaert et al. 1940) which are both only in printed form available. The Sun and Arcturus are of spectral types G2 and K2 respectively; Procyon is an F5 type star. This Atlas is based upon photographic spectrograms obtained at the coude' spectrograph of the Mount Wilson 100-inch reflector during three observing seasons. The instrumental profile is carefully described and illustrated in the Introduction to the Arcturus Atlas, and applies equally to the present Atlas. Numerical representations of the central part of the profile at particular wavelengths in the second and third orders of diffraction have been published elsewhere (Mackle et al., 1975). Twenty-eight spectrograms exposed on the Eastman Kodak emulsion IIa, IIa-D, IIa-F and 103a have been used in the production of the Atlas. The calibration was done by exposing the plates with either a wedge slit or a stepped entrance aperture. It provided a calibration curve spanning 1.7 logarithmic units and defined at intervals of 0.1 unit. The various slit widths were distributed at random, in an effort to avoid systematic effects. The wavelength scale was linearized by interpolation. The wavelengths used were either those resulting from actual measurement of the complete plate, or failing that were solar wavelengths adopted from the catalogue by Pierce and Breckinridge (1973). Individual lines typically give residuals of the order of 3 milli-Angstroem (=0.0003 nm), and in parts of the Procyon spectrum which have many good lines the wavelength scale is expected to be accurate to this order of precision. Inevitably, there may be greater errors in the barren regions of the spectrum which occur at longer wavelengths. The continuous spectrum of Procyon appears to be visible throughout the region covered by this Atlas, at least in the sense that there are frequent high points lying on, or close to, a common locus. This continuum has been drawn by eye on the raw intensity plots, and then normalized to a uniform value of 100. A substantial difficulty arises with the hydrogen lines and Ca II H and K owing to their great extent: the lower members of the Balmer series cover something like 10.0 nm, and the wings of the higher members overlap to such an extent that the lines effectively merge into a pseudo-continuum at a wavelength (about 368.0 nm) still well above the series limit of 364.6 nm. In the wavelength region of 364.6-400.0 nm we have adopted, for the purposes of the Atlas, an operational definition of the continuum as being the locus of the maxima of the _local_ continuum drawn in the overlapping wings of the Balmer lines. In the near-ultraviolet, therefore, the true continuum must lie far above the Atlas continuum, and moreover it has a large discontinuity (the Balmer Jump) at 364.6 nm. Proceeding shortwards from the red, the true continuum probably begins to diverge from the Atlas value near 400.0 nm, if not before, and to rise increasingly steeply to a level of perhaps 150-200 per cent in the vicinity of 368.0 nm, a level from which it returns stepwise to the Atlas continuum at 364.6 nm. To obtain a good continuum definition near the hydrogen and H and K lines, a spectrum with a very compressed wavelength scale have been used. The bridging process is easily accomplished where the lines are comparatively close together, but difficult for the first four Balmer lines where the distance to be bridged is of the order of 10.0 nm, corresponding to 6-10 cm of plate, and for this purpose very high dispersion is a distinct disadvantage. The first few pages in this Atlas cover a region of the spectrum where telluric ozone absorption is increasing rapidly towards shorter wavelengths -- indeed, for ground-based observers the absorption becomes effectively infinite only 20.0 nm beyond the shortward limit of the Atlas. The onset of ozone absorption is not smooth, but seems to include bands, typically several Angstroems (0.1nm) wide and of irregular profiles, where the absorption is several per cent greater than would be expected on the basis of interpolation on a smooth curve. The bands are too narrow, and the data on their character too scarce, for the continuum to be corrected for their presence: the continuous spectrum used in normalizing the Atlas tracings has simply been bridged, in this region just as elsewhere, between local maxima in the observed intensities. Band absorption by ozone is probably significant for present purposes only below 330.0 nm.