On errors arising in the measurement unsymmetrical spectrum lines
It is generally recognised that one of the most serious sources of error in relative determinations of the wave-lengths of spectrum lines by micrometric measurements of the positions of the lines on photographic plates lies in the mechanical and other displacements, which result in errors in the superposition or juxtaposition of the spectrum under investigation and the standard spectrum, the wave-lengths of which are supposed to be known. Thus, if a spectrum to be measured is photographed on a plate and an iron arc spectrum is then photographed in juxtaposition, an error may arise, either from some flexure or mechanical displacement between the two exposures or from a gradual change of temperature of the dispersing prism or grating, resulting in a continuous drift of the spectrum across the plate. Even when the utmost care is taken to avoid these displacements by rigid design of the spectrograph and proper control of temperature, it is hardly ever possible to eliminate them entirely, and it is necessary to follow some procedure which will reduce the errors arising from them to a minimum. A method which is commonly used is to expose the comparison spectrum in two parts. For example, a comparison spectrum with one-half the total exposure required may be taken first; the spectrum to be measured is then exposed for an appropriate time, and finally the second half-exposure of the comparison spectrum is superposed on the first. Images of the comparison lines (and in the case of a long exposure of the lines to be measured) would be somewhat broadened, but it might at first seem that the maxima of the lines in the two spectra would be in correct juxtaposition, supposing that there has been a constant drift of the spectrum across the plate owing to a gradual change of temperature. St. John and Babcock, in a study of the pole effect in the iron arc, used a system of reflecting prisms to project on theslit simultaneously the light from the neighbourhood of the pole in juxtaposition with that from the centre of the arc, the intensities being equalised by means of a rotating sector, which occulted for an appropriate period the more intense light from the neighbourhood of the pole.