'In order... to test the validity of the prevalently-accepted theory [that the arrangement of eye-muscles is the most appropriate means of explaining the tendency to overestimate the lengths of horizontal compared to vertical lines], a series of experiments was [sic] devised and carried out, during the session 1903-4 and following sessions, in the Cambridge Psychological Laboratory. These experiments consisted in studying, by means of the method of right and wrong cases, the way in which the illusion was affected by the durtation of exposure. A series of momentary exposures of the figure, short enough to exclude the possibility of any such eye movements as as would be necessary if estimation of length were dependent on them, and a like series of prolonged exposures were given, and after each separate exposure the observer recorded his judgement, which was taken down by the experimenter.

The method employed was to exhibit a vertical and a horizontal line at right angles to one another. The horizontal line was to the right of, and below the point of junction with, the vertical line, so that the figure resembled the letter L. The length of either line could be varied at will. The figure was projected on a screen by means of the apparatus shown in Fig. 1 [note: 'This apparatus was devised by Mr F.N. Hales.']. The rays of light from an incandescent gas-lamp, leaving the condenser A, pass through a rectangular slit in the screen B shown on a larger scale in Fig. 2. The whole aperture in the screen is in the form of a cross of which any part or the whole of each limb can be closed by small shutters which slide in grooves. The sides of the grooves are graduated so that the length of each limb can be recorded. In Fig. 2 the slit is shown as it was used in our observations, an inverted image being thrown on the screen. The angular beam of light from the slit passes through an aperture in the pendulum C and then by means of the lens E an image of the figure is thrown on a translucent glass plate in the screen F. When no obstacle is interposed an observer sees on the screen F an image which has half the dimensions of the slit in B. In order to obtain brief exposures, the pendulum C is placed so that it is held by the magnet, being released when the key G is opened and caught at the end of its swing by the catch D. The time of exposure can be regulated by varying the size of the apeture in the pendulim and the distance through which the pendulum swings. In the observations we call 'momentary' throughout this paper the exposure was less than 1/50th of a second and was probably short enough to exclude all movements of the eyes during the exposure, and was certainly sufficient to exclude all movements along the two lines of the figure as would be necessary on the supposition that estimation of their length is based on muscular sensations.' (245-246)