84 



DESIGN AND USE OF INSTRUMENTS 



curve of best fit. Craik chose a simple 

 function but a step-shaped curve would 

 probably have fitted his data as well. And 

 second, does the step in the Spragg and 

 Rock data reflect some change in the na- 

 ture of the reading performance? Did 

 the subjects experience a change, for ex- 

 ample, from reading under conditions of 

 "good seeing" to reading under conditions 

 where they could not discern the individual 

 scale marks or see the numbers clearly, but 

 where they could still infer the approximate 

 value associated with a given pointer setting 

 because of their familiarity with the scale 

 and its numbering? The error records 

 shown in Part B of Fig. 5 reveal that the 

 subjects' accuracy changed as abruptly as 

 their reading speed at the critical brightness 

 level. This occurrence of corresponding 

 error and speed changes suggests again the 

 desirability of knowing more about perform- 

 ance in this region. A general reconcihation 

 of these results with those of Craik will be 

 welcomed. New studies may well be con- 

 cerned also with the question of why the 

 critical brightness levels in the two sets of 

 results were so widely different. 



It is of interest to observe the specific 

 points plotted for red and white illumination 

 in Craik 's graph. Fig. 4. With the excep- 

 tion of readings at the lowest intensity of 

 red light, reading under red and white light 

 was equally rapid. These data are in fine 

 with the results of Spragg and Rock's filter 

 study discussed above. 



Reference may be made again at this 

 point to Fig. 3 dealing with the effect of in- 

 tensity of illumination on reading speed for 

 instrument numbers of different size. One 

 partial check on the adequacy of these data 

 is found in the results of Lee and Finch (49), 

 who determined what the brightness of red 

 illuminated instrument numbers should be 

 in order for them to be clearly legible. For 

 numbers 18 minutes by 14 minutes in angu- 

 lar size, the required brightness was 0.009 

 foot lamberts. This value agrees satisfac- 

 torily with Craik's determinations for 15- 

 minute numbers. 



Instruments for Presenting 

 Quantitative Data 



Counters vs. Scales 



For indicating precise numerical informa- 

 tion, a counter type of instrument can prob- 

 ably not be surpassed. If such an instru- 

 ment is well engineered, the reader has no 

 interpolation problem. He simply reads the 

 digits of the number as they appear in the 

 counter aperture. Design problems reduce 

 to those of making the numbers legible and 

 spacing them properly. 



Comparative data on the relative merits 

 of scales and counters for the presentation 

 of quantitative information can be found in 

 Grether's study of altimeter designs (31). 

 The counter was read faster and with fewer 

 gross errors, of 1,000 feet or more, than the 

 simplest scaled instrument tested, a straight 

 vertical scale. (See Fig. 11 below.) Sim- 

 ilar results were obtained by Chapanis (14). 

 As the expression "direct reading counter" 

 implies, a counter requires essentially no 

 interpretation. This makes the counter a 

 desirable indicator in situations where speci- 

 fic quantitative data are to be read. 



There are situations, however, in which 

 the counter is not the most satisfactory form 

 of indicator. One of these is for the task 

 of instrument setting or adjustment. An 

 operator who is trying to come to a required 

 setting using a counter sees nothing but a 

 whirl of digits. Thus, the counter developed 

 by Riggs for field artillery apphcation (9, 66) 

 and another by Chapanis for presenting 

 radar scope bearing data (15) were recom- 

 mended for use in reading bearing informa- 

 tion but not in making bearing settings. 

 For setting operations, dials or scales are by 

 test more satisfactory. 



Scaled instruments, of course, have other 

 advantages pecuhar to themselves which 

 make them useful for many functions. 

 They can often be set up in such a way as to 

 be semi-pictorial. They can also be de- 

 signed in such a way that makes for rapid 

 check reading when precise quantitative data 

 are not needed. Thus, the scaled instru- 



