Gratings similar to those described and studied by the author in 1907 and 1912 have been more fully investigated. They show narrow bright and dark bands in the continuous spectrum of a white source. Classical theory has never accounted for these anomalies, though tentative efforts were made at the time by Lord Rayleigh. The present work was done with a chromium plated echelette grating of 7200 lines to the inch ruled on copper, and one of 15,000 lines ruled by Dr. Babcock on an aluminum film on glass. The behavior of the bands as the incidence angle is altered has been very completely recorded by photography and the energy missing in the grating spectra has been found in excess in the spectrum of the central image. The dark bands appear to be due to the circumstance that diffracted wavelets from the lines are inhibited by the collective effects from neighboring lines, i.e., there is a sort of destructive interference along the plane of the grating. This obtains, however, only when the grating space is equal to an integral multiple of $\lambda$ (for normal incidence) which means that the dark bands correspond to $\lambda$ values which are passing off the grating on both sides at grazing emergence. The spectra formed by the Al grating are completely plane polarized, because of the circumstance that the width of the scratches is considerably less than the wave-length of the light. The dynamical action of the gratings is discussed, and the anomalous behavior is shown to be due to the presence of exceedingly narrow diffracting elements, which may be present even in the case of rather coarse rulings.

• Received 11 July 1935

DOI:https://doi.org/10.1103/PhysRev.48.928