NTJMERA.LS 



69 



this operation is even more analytic than 

 that of reading them upright. 



Type Face 



Tinker (48) found that modern and old 

 style digits were read with equal speed and 

 accuracy under normal reading conditions. 

 When presented in groups at a distance, the 

 old style digits were easier to perceive. As 

 the two sets are not very different in details 

 of form, the author attributed the advantage 

 of the old style to the fact that some of the 

 digits in this set project above or below the 

 line. However, adjacent rising digits tended 

 to confuse each other, as Roethlein (38) 

 found for letters. 



White on Black 



It will be recalled that, for verbal material, 

 white on black is less legible than black on 

 white. In the Tufts study with printed nu- 

 merals (16), white on black was read with the 

 same speed and accuracy as black on white. 

 Dunlap (22), in an investigation of license 

 plates, concluded that dark numbers on a 

 light ground were better, but this result, 

 according to Berger (3) was probably a func- 

 tion of stroke widths used in the study. 

 Berger 's own data show that w^hite license- 

 plate numbers on a dark ground are more 

 perceptible than black on white when each 

 set is prepared in its optimal stroke width. 



Digit Configuration 



The present designs of the digits, like those 

 of letters, were determined to a large extent 

 by convenience in making the strokes before 

 printing was invented. With printing, type 

 makers have modified numeral details, per- 

 haps more often for aesthetic reasons than 

 for ease of discrimination. As a result, in 

 some type faces the three and five are differ- 

 entiated by one small vertical segment only. 

 Sixes and nines are sometimes made like 

 zeros with small tails, and are difficult to 

 identify under near threshold conditions. 

 Work by Tinker (44) and by Berger (3) 

 shows that the legibility of the digits varies 



considerably. If our conventional number 

 forms could be disregarded, it is probable 

 that a much improved set of symbols could 

 be devised. Short of that, we have Berger 's 

 (3) work in improving the configuration of 

 license-plate numerals, in which he found the 

 angles of the slanting fines to be one im- 

 portant factor. Work on this general prob- 

 lem with dial-type digits is under way at 

 Tufts (19). 



Spacing 



The legibility of digits presented in groups 

 is different in both relative and absolute 

 terms from that of digits presented singly 

 (Tinker, 48; Berger, 3). Berger found that 

 spacing could be used to some extent to con- 

 trol the thresholds. It is possible, therefore, 

 that printed numerals which usually appear 

 in groups may require different designs from 

 dial numerals which often appear singly or 

 in limited combinations. 



For further discussion of the influence of 

 some of these variables on numeral legibility, 

 the reader is referred to the section in 

 Chapter 3 which deals with instrument 

 marking. 



Other Factors in Digit Legibility 



Illumination 



Atkins (2) found that time scores on a 

 number cross-out test differed by less than 

 four percent for five illuminations ranging 

 from 9.6 to 1 18.0 foot-candles. In the Tufts 

 studies (17), time and error scores in reading 

 printed numerals were not significantly 

 different under brightnesses from one to 15 

 foot-lamberts. (Again, see Chapter 3 for 

 related data on instrument dial numerals.) 



Vibration 



In the Tufts experiments (16, 17, 18, 19) 



vibration was introduced into the visual field 

 by optical means. With secondary condi- 

 tions moderately severe, a double amplitude 

 of vibration of 0.02 inch was sometimes 

 found to produce a measureable impairment 

 in time or error scores. The effect varied 



