68 



PRINTED MATERIALS, MAPS, CHARTS 



Tinker (57), using the same type faces pre- 

 viously tested for reading speed, found sub- 

 stantial differences in distance thresholds. 

 American Typewriter, especially poor for 

 reading speed, was especially good by the 

 distance criterion. They attributed the dis- 

 tance results to such factors as relative size 

 and boldness, simphcity versus complexity of 

 design, area of w^hite spaces, and relative de- 

 pendence on hair lines. 



Such comparisons suggest that the reading 

 of continuous text is a highly integrative 

 process to which context, word form, and 

 detail of letter design all contribute; that the 

 recognition of words at a distance is less 

 integrative in that it makes less use of con- 

 text; and that the recognition of letters at a 

 distance is the most analytic of the three, 

 being dependent entirely on detail of design. 



Numerals 



The reading of printed numerals under 

 ordinary working conditions stands at the 

 analytic end of the continuum set up in the 

 preceding section. Context and word-form 

 cues are lacking, and correct identification of 

 each digit is usually critical. Integrative 

 habits which enhance efficiency in reading 

 verbal text may be disadvantageous with 

 numerals. For example, it has been found 

 that familiar number combinations such as 

 1492 tend to be read hke word units (Rebert, 

 37). This tendency could cause a combina- 

 tion similar to a familiar one to be misread. 



It could be expected that the more analytic 

 the reader's function, the better the chance 

 of predicting relative legibility from knowl- 

 edge of the visual mechanisms, and also that 

 the nearer to a threshold in any respect, the 

 better the chance of such predictions. Dial 

 numerals under night illumination constitute 

 the extreme example. (See Chapter 3 

 below.) Among the pointers and leads in 

 this connection, we have Craik's (15) data 

 on the relation between numeral size and 

 illumination for constant legibility under 

 night conditions, Berger's (3) experimentally 

 verified prediction from acuity data, that the 



optimum stroke width would be greater for 

 black digits on a white ground than for the 

 reverse, and Sanford's (40) earlier analysis 

 of the design details that differentiate the 

 lower case letters. Such devices as the Luc- 

 kiesh-Moss "visibility" meter, and the Weiss 

 focal variator (58), w^hich gives thresholds in 

 terms of impairment of focus, might prove 

 more helpful for numeral legibility than for 

 verbal text. A survey of the sensory factors 

 in digit design has been made for the Special 

 Devices Center by Gleason (23). He con- 

 cluded that such factors as size of critical 

 detail, brightness, and contrast are well 

 enough understood, but that more work is 

 needed on stroke width and configuration. 



The helpfulness of general theory does not 

 emancipate us from empirical checks, espe- 

 cially where complex interactions may play a 

 part. No integrated program on numeral 

 legibility comparable to that of Paterson and 

 Tinker on typography in ordinary reading 

 has been carried out, but scattered studies 

 on various aspects of the problem can be 

 cited. 



Typographical Factors in Digit Legibility 



Type Size 



Type size is one of the factors being in- 

 vestigated by Crook, Hoffman, Kennedy, 

 and associates in a series of studies on nu- 

 meral legibility currently in progress at Tufts 

 College (16, 17, 18, 19). In some experi- 

 ments the reading task has involved a se- 

 quence of "same-diiferent" judgments on 

 pairs of digits. In others it involved a se- 

 quence of mental addition problems. One- 

 to three-minute trials were separated by 

 comparable rest intervals. The material 

 was printed in Modern type face. With the 

 numerals in the normal working position, 

 type sizes from 6- to 11-point were not differ- 

 entiated in time or error scores. With the 

 material inverted, however, significant differ- 

 ences appeared in both measures, small type 

 being read less well (17). The disadvantage 

 of small type for reading the digits inverted 

 suggests, in the terms proposed above, that 



