80 



DESIGN AND USE OF INSTRUMENTS 



forms which he finally developed are quite 

 different from those in general use on in- 

 struments or in printed copy. Before being 

 adopted for general use, they must, of course, 

 be evaluated for other sizes and reading 

 distances, and by criteria other than the 

 one which Berger used. Important, how- 

 ever, for the present discussion is the fact 

 that Berger's numeral shapes were empiri- 

 cally determined. It was on the basis of 

 experimental evidence, and not just a series 

 of "hunches," that he decided (1) to make 

 the angle at the base of the 2, in the 7, and 

 in the tails of the 6 and 9 as near to 45 de- 

 grees as possible, (2) to close the angle at 

 the top of the 4, (3) to run the bar on the 

 5 over the entire number, and (4) to make 

 the center point on the 3 midway fall be- 

 tween right and left borders of the figure. 

 Some agreement with Berger's designs, par- 

 ticularly for the numbers 2, 4 and 9 is found 

 in studies of time-table digits which Fish- 

 enden (26) reviews very hastily. 



The numbering of most service instru- 

 ments in this country today follows a mod- 

 ified Le Roy lettering style, like that shown 

 in Part B of Fig. 1. Changes introduced 

 in this standard style from time to time 

 appear to have been made principally on a 

 priori grounds (93). Currently, the recom- 

 mended ratio of stroke thickness to numeral 

 height in this series is 1:6. 



British work on numeral designs seems 

 to have been limited to that of Craik (19). 

 He determined the legibility of instrument 

 numbering by observing the speed with 

 which trained subjects read numbers at dif- 

 ferent brightness levels. He summarized his 

 results by saying that at any given digit 

 height, legibility was greatest when the white 

 parts of the figures were equal in breadth 

 to the mean width of the enclosed black 

 parts. This can be interpreted as signify- 

 ing a ratio of stroke width to digit height 

 of 1:5. Craik's number series is shown in 

 Part C of Fig. 1. Confusion between the 

 6, 8, and 9 was said to be less hkely when 

 the digits were of the form shown. 



Stroke thickness as such as been the sub- 

 ject of several other studies. An early series 

 of tests by Paterson, Walsh, and Higgins 

 (60) on the legibility of self-luminous nu- 

 merals pointed, as did Berger's later data on 

 luminous numerals, to the desirability of 

 low stroke-\%'idth to digit-height ratios. 

 More recently, Loucks has compared the 

 readabihty of instruments marked with num- 

 bers of different degrees of boldness. In 

 the first experiment (51), he compared num- 

 bers 0.19 inches high with stroke widths 

 of about l/6th and l/4th letter height. 

 Under ultra-violet illumination, at what is 

 described as a relatively intense level, dials 

 numbered wdth these two different sets of 

 digits were read in tachistoscopic exposure 

 with equal accuracy. Under a higher tung- 

 sten illumination, however, the dials with 

 the 1:6 numerals were read with signifi- 

 cantly fewer errors. In a companion study 

 of numeral markings on climb and dive in- 

 dicators (52), comparable results were ob- 

 tained. In ultra-violet illumination dials 

 with 1:16 numerals were read less accurately 

 than dials with 1:8 or 1:5 numerals. In 

 tungsten illumination, however, the 1 : 16 and 

 1:8 numerals were equally satisfactory and 

 superior to the 1:5. These data are con- 

 sistent in suggesting that the optimum bold- 

 ness of white on black figures decreases as 

 illumination intensity increases. 



Digit Proportions. The problem of nu- 

 meral proportions, the ratio of numeral width 

 to height has received Uttle, if any, experi- 

 mental attention. Loucks (51), in discuss- 

 ing possible extensions of his research, sug- 

 gests the need for legibihty studies with 

 numbers of proportions different from the 

 ones used on his test scales. He states that 

 "tall but narrow overall width figures . . . 

 have been found to give maximum legibihty 

 in combination with minimal interference 

 with night vision." No reference for the 

 latter comment is given, however. Such 

 a change in numeral proportions corresponds 

 to the change made in going from normal 

 to "condensed" type. In as much as evi- 



