For the purpose of our investigation and other investigations at the Nutrition Laboratory, it seemed highly desirable to eliminate this criterion for judgment. The iris diaphragm (D, fig. 39) successfully solved the problem. In the apparatus as used the test object is located 170 cm. from the eye of the subject. The diaphragm D is 31 cm. from the subject's eye. Thus, when he accommodates for the distance of the test object, the edge of the diaphragm is very hazy and indefinite. By reducing the opening in the diaphragm to a diameter of 12 nun. and properly placing the tube and the artificial pupil in relation to the diaphragm and the test object, the subject was unable to see any of the circular frame surrounding the test field and could not see the entire field. The portion exposed to view was circular, with indefinite edges and approximately 7.5 cm. in diameter. This reduced to a negligible quantity any disturbance from the phenomenon illustrated in figure 51, but with the definite boundary of the test field hidden from view, and in its place a hazy fading-out of the lines on all sides, the subject found difficulty in keeping the eye accommodated to the proper distance when lines were not in view, as just preceding the taking of threshold measurements. It was found possible to fasten a small disk of black paper 2 mm. in diameter to the surface of the glass grating which was nearer to the subject. This provided a fixation point for accommodation before the lines of the field came into view. The central line of the field directly behind the fixation point changed only in width, and a few lines on either side had but slight lateral movements. This position was, therefore, very favorable for fixation. The artificial pupil performed an important service in making a fixed view-ing-point for the subject. Had it been larger, or had there been no artificial pupil, it would have been possible for him, by moving to one side or the other, to have exposed the definite, circular frame of the test field. With a 3-mm. opening of the artificial pupil, it was impossible to get in any position with the apparatus as used, from which the definite edge of the test field could be seen, and when the subject moved slightly one way or the other, he could immediately recognize the dimming of the field and that he must be out of position.

1 Ives, Abstract Bulletin of the Physical Laboratory of the National Electric Lamp Asso., 1913, 1, opp. p. 36.

The luminant was an 8 c. p. carbon filament lamp. The brilliancy of the test-object window in candles per square meter of surface was 20.3. With a 16 c. p. carbon lamp, the brilliancy in candles per square meter was 57.8. In Cobb's apparatus the diameter of the window exposed to view was 3.5 cm. The brilliancy of the source in candles per square meter he varied from 5.94 to 189.0. Increased intensity of illumination beyond a certain point does not greatly aid the eye in distinguishing detail. Fig. 40 (p. 170), which gives results with 8 c. p. and 16 c. p. lamps, shows that although the latter supplied an illumination about three times that with the 8 c. p. lamp, yet the visual efficiency was not greatly improved. As the brighter light was much more tiresome to the eye, the 8 c. p. lamp was used in illuminating the test object for these experiments.

The first time a subject was given this test at the Laboratory he was fawtnwted as follows:

"You will now look through this small peep-sight with the right eye and see a light window about 6 feet away from you. In the window are dark bands.Please notice that they may be made small or large as we desire. We want to discover the are of the smallest lines that you can see. Our method is as follows: We will first make the lines so small that they disappear from view; they will then be gradually widened until you can see them, whereupon you will call 'stop.' Please notice, also, that the lines may be vertical or bonsontal or they may be in a diagonal (45°) position with their tops to the right or to the left. The four positions will be used, viz, right, vertical, left, and horizontal. You can never tell beforehand in what direction the lines are going to appear, because the lines are made too small to see, and the instrument is set for some one of these four directions in a chance order. Do not try to anticipate the direction in which the lines will come. When you are given the signal to be ready, look intently at the black dot in the middle of the field. The instant you know the direction of the lines call 'stop' and give the direction. After you have done this, rest the eye by closing it until the next' ready' signal".

The right eye of the subject was always used for the determination; it was not refracted and glasses were never worn. In these experiments it was not the object to attain absolute values or to reveal individual differences, but to test the man, for comparative purposes, against himself in successive sessions. Since the eye was not refracted it is obvious that astigmatism would play a r61e in influencing readings in certain axes. The time at our disposal for taking the measurements on a single subject was approximately 10 minutes. In this interval it was usually possible to make from 3 to 5 separate threshold determinations for each of the four positions, right, vertical, left, and horizontal. These followed each other in a chance order. After two or three trials the experimenter knew, approximately, where to expect threshold value. The micrometer wheel was, after this, advanced, at first one turn in about 5 seconds, and then the rest of the way to threshold at the rate of one revolution in 15 seconds, as the object was to get the reading within an interval of less than a minute after the subject had been given the "ready" signal and fixated on the black dot in the middle of the test field. A comparison of results from this apparatus, with standards previously established, will be considered in the discussion of our results (see page 607) for the reduced diet experiment.

It is most useful to state the threshold-test band width for any subject in terms of degrees on the arc of vision. In the Nutrition Laboratory instrument the distance from tip to tip of the yoke Y is 119 mm. The separation of the ruled lines on the glass plates is 1/240 inch or 0.10577 mm. According to the formula given by Behn, Ives, and Johnson, in the articles cited, the constant for this instrument is therefore 1.573. This factor divided by the micrometer scale reading on the instrument gives in millimeters the width of the separation of the light bands. To illustrate in another way, if the micrometer wheel is set to read 1.57, measurement by a millimeter scale applied directly to the window of the test object will show that both the light and dark bands are each 1 mm. in width, i. e., from the center of one dark band to the center of the adjoining light band is 1 mm. A scale-reading of 3.14 will show bands 0.5 mm. in width, and so on. In the arrangement employed the window of the test object is distant 170 cm. from the subject's eye. On the circumference of a circle with this radius an arc of 1 second = 0.008242 mm. The millimeter width of test bands just distinguishable to the subject's eye may be divided by 0.008242 to state the threshold in seconds on the arc of vision. For example, a micrometer-scale reading of 3.14 shows bands 0.5 mm. in width. This width of 0.5 mm. represents to the subject's vision an angular separation of 60.7 seconds between the bright lines of the test field.1