A determination of the spectral distribution of the solar radiation is given on the basis of observations of the solar continuum and the absorption by Fraunhofer lines. The ultraviolet and infrared correction suggested by Schatzmand an Allen and those adopted by Abbot are compared. The mean value if the solar constant should be corrected to allow for the exact spectral distribution between 1u and 2u, where the more important correction appears. If a value of 1.94 cal-min-¹cm-² is assumed for the standard value of the Smithsonian Institution, a new estimate of the solar constant gives 1.98 cal-min-¹cm-². However an absolute value must be given by new absolute measurements of the solar emission in certain spectral ranges or by direct measurement by rockets. The error of the solar constant is probably not less than ± 5%.