IN ITS RELATION TO THE EYE. 467 



reflectors in the field of view. This reflector is designated in the 

 charts as Reflector III. All of the reflectors of this series were in- 

 stalled on the ceiling pendant in accord with the principles of direct 

 lighting. 



It was our wish to conduct this investigation, as has been the 

 case in all of our work on the distribution factors, with the color 

 value and the intensity of light as nearly equal as possible at the test 

 object. Clear tungsten lamps were used with Reflectors I., II. and 

 IV. ; full-frosted lamps with Reflectors VI. and VII. ; and bowl- 

 frosted lamps with Reflectors III. and V. Clear lamps were used 

 in the cases mentioned because in the first place in reflectors of this 

 type the lamps were not visible to the observer at the point of work; 

 and secondly, although the illumination given was high in the aver- 

 age, the distribution was such as to give a low illumination at the 

 point of work. That is, the tendency of these reflectors, installed at 

 the height used in our test room, was to give lanes of light directly 

 beneath the two rows of reflectors, shading off to a correspondingly 

 low value on either side. Full frosted lamps were used with Re- 

 flectors VI. and VII., because with Reflector VI. a part and with 

 Reflector VII. all of the filament would otherwise have been visible 

 to the observer; also the value of the illumination at the test object 

 would have been much too high as compared with the other reflectors 

 in the series and higher than the values used in previous work. In 

 case of Reflectors III. and V. both of the above objects, namely, the 

 better protection of the eye from the filament and the performance 

 of the tests with the illumination values as nearly as possible equal 

 to those obtained with the other reflectors and in previous work, was 

 best accomplished by the use of bowl frosted lamps. 



The results of this series of tests are represented in Chart VII. 

 Again in order to ascertain which of the brightness specifications — 

 order of magnitude of highest brilliancy; average brilliancy; ratio of 

 highest to lowest order of brilliancy ; ratio of highest order of bril- 

 liancy to average brilliancy ; ratio of average to lowest brilliancy ; 

 ratio of highest order of brilliancy to brightness at point of work 

 (brightness of test object and reading page) ; etc. — correlate most 

 closely with the results for the tendency to cause loss of power to 

 sustain clear seeing, charts were constructed in which a number of 



