METHODS OF RESEARCH ON VISIBILITY 



105 



small rectangle of the phosphor screen elec- 

 tronically excited at controllable inter- 

 vals and intensities. Though little used as 

 yet in visual studies, it was successfully 

 employed by Bartlett (3) to study flash 

 frequencies and by Hanes (18) to study 

 comparative brightness judgments. More 

 commonly, the viewer is a photocell or a 

 phototube. However, data can be corre- 

 lated with knowledge of visual requirements 

 The correlation is more often implicit than 

 exphcit and is incHned to be based on esti- 

 mates of a "photopic eye." 



Optical Simulation 



The work done in England by Hopkinson 

 (24) employed only an optical imitation of a 

 radar scope. Such studies require an as- 

 sumption of successful imitation. Often this 

 can be plausible. Some of the optical imita- 

 tions developed for training purposes may 

 even be adaptable to research (33, 36, 47). 

 Data from imitations are yet to be compared 

 with those obtained from real scopes. 



Physical Measurements 



The critical stimulus for vision is difficult 

 to measure, especially on intensity modu- 

 lated scopes. In A-scopes, the deflection is 

 discriminated by its relative height and by 

 the width of the gap in its baseline. A psy- 

 chophysical study of each cue would require 

 their independent variation, a nearly impos- 

 sible task on a real scope. The practice has 

 been to measure the voltage or power of the 

 deflection, rather than its extent, relative to 

 some selected "average" noise voltage or 

 power. From a purely visual standpoint the 

 threshold A-scan deflection is some mongrel 

 form of a visual acuity, and the signal-to- 

 noise ratio is meaningless. In intensity- 

 modulated scopes, detection requires a 

 combination of form and brightness discrimi- 

 nation, and the signal-to-noise power is an 

 equally inappropriate unit. For brightness 

 discrimination experiments one would like to 

 use brightness units. Because of variations 

 in electrical circuits and because of the 



marked effects of afterglow (in the memory 

 screens) on brightness level, it is virtually 

 impossible to control screen brightness pre- 

 cisely by means of meters or scopes that 

 register only volts or amperes. Measure- 

 ments made at Systems Research Labora- 

 tory, for example, showed a difference of 

 about seven volts for the visual cut-off bias 

 between an old tube and a new tube of 

 identical type. Over months of daily use, 

 this bias in the new tube gradually receded 

 in the direction of the old tube. This means 

 that it is never safe to compare tubes or con- 

 ditions by means of a voltmeter alone. It 

 became necessary for the workers at Systems 

 Research Laboratory to depend on a visual 

 estimate of sweep cut-off as a reference point 

 for determining background brightness (55) . 

 Even using a visual reference, brightness 

 differences must be measured by meters, for 

 subjective estimates of supra-threshold 

 brightness levels are not very accurate. It 

 would be desirable to have a physical instru- 

 ment sufficiently adaptable to use continu- 

 ously as a monitor for the brightness of the 

 scope, but such instruments are not availa- 

 ble. The closest approximation is an elec- 

 tron multiplier tube, (usually either an RCA 

 931 A or 1P22 multipher phototube) in con- 

 junction with a dc amplifier. These tubes 

 can amplify weak photo-currents as much as 

 200,000 times, but they lack two desirable 

 features: their lower limit of Hght sensitivity 

 is between 0.01 and 0.1 footlambert, which is 

 still about three log units above the eye's 

 threshold; they are likely to be cumbersome 

 to focus and to read, without extremely 

 elaborate and expensive accessories. 



In the absence of a brightness scale, 

 thresholds of signals have conventionally 

 been expressed in decibels of attenuation of a 

 reference signal voltage or power, assuming 

 constant noise. The scale is arbitrary with 

 respect to absolute values of energy but is 

 entirely meaningful for comparative pur- 

 poses. Inasmuch as satisfactory agreement 

 has never been reached on the definition of 

 noise, different investigators select different 



