A simple machine is described which quickly and automatically records on a chart the wave‐number intervals between each line in a spectrum and every other line within any desired range. Small rectangular holes are punched in a black paper tape, their positions corresponding to the wave numbers of the lines on a linear scale. This operation is rapidly carried out by means of a magnetically operated punch, working on the tape which passes over a cylinder from which wave numbers can be directly read off. The punched tape is then fed from a reel through a simple device which causes it to return on itself after passing for one meter directly over a sheet of bromide paper which is moving at right angles to the tape at a fraction of the speed of the latter. The returning tape, after passing over itself for a meter, is pulled between two rollers and stored on a reel. A set of lights is hung over the doubled tape, and whenever two punch marks come into coincidence a small rectangle on the photographic paper is exposed. Fairly high tape speeds can be used, so that a complete spectrum can be recorded in a few moments. On developing the record, the photographic paper, 40 inches wide and as long as needed, is found covered with small rectangular dots; the abscissa of each rectangle gives the interval between the two dots producing it, while its ordinate gives the actual wave number of either. Suitable marks having been punched into the tape before and after the spectrum, the calibration of the interval and wave‐number scales to compensate for any shrinkage of the record is automatically made. Those lines having any given separation in the spectrum can be quite readily picked out from the multitude of dots by means of a straight‐edge. By suitably varying the scale used for plotting wave numbers on the tape, the width of the punch marks, and the ratio of tape speed to paper speed, each type of spectrum can be treated in the most suitable manner. Where very large intervals must be recorded while a fairly open scale is still used arrangement is made for keeping a desired constant amount of slack in the returning tape, so that on one record all intervals between 30 and 2030 cm−1 might appear, on another 2000 to 4000, etc. Provision can also be made for determining from the appearance of a dot the approximate intensities of the lines combining to produce it.