INSTRUMENTS FOR PRESENTING QUANTITATIVE DATA 



89 



they, like the first group, make significantly 

 fewer reversal errors on the new scales. In 

 the course of this experiment both clockwise 

 and counterclockwise scales were used. 

 Since Christensen did not analyze his data 

 to observe the effect of the staircase on each 

 direction of scale separately, it is possible 

 that further useful information could be ob- 

 tained from a re-analysis of his records. 

 Further study of this method of scaling will 

 l^e anticipated with interest. 



Scheme for Numbering Graduation Maries. 

 Some results obtained by Vernon (79) link 

 scale numbering with the graduation prob- 

 lem. Wherever possible, it appears desir- 

 able to have different scales which are to be 

 read in succession designed with the same 

 numbering and the same graduation values. 

 Vernon observes, however, that if graduation 

 interval values must be different for scales 

 which are to be read successively, less con- 

 fusion is caused by using the same number 

 series on the scales and varying the intervals 

 between them than by keeping the number 

 of divisions between numbers constant on all 

 scales. This problem deserves reinvestiga- 

 tion because it is not clear to what extent 

 Vernon's conclu.sion, from admittedly pre- 

 liminary data, resulted from the exception- 

 ally confusing graduation values used on 

 some of her scales. 



Without offering specific supporting data, 

 Vernon advances the following principles or 

 rules of scale numbering: (1) that numbers 

 on the scale should not be so close together 

 as to cause confusion, (2) that large numbers 

 are to be avoided, two-digit numbers being- 

 better than three, (3) that decimals are to be 

 avoided, (4) that numbering by tens (or tens 

 multiples thereof) is most desirable, (5) that 

 the maximum, and probably the optimum, 

 nimiber of divisions between scale numbers 

 is ten. These rules appear to have been 

 reasonably well recognized by scale designers 

 in the past. 



The only reported experimental inquiries 

 into the ciuestion of the number of numl)ered 

 graduations to have on dials which are actu- 



ally in service use are those of Loucks (50, 

 51, 52, 53). Loucks studied various de- 

 signs of tachometer, climb and dive indicator 

 and manifold pressure dials. His procedure 

 Avas to expose individual dials for short in- 



Fig. 10. Examples of pairs of dials tested by 

 Loucks (50, 53) 



tervals, usually 0.75 seconds. He used 

 tungsten, red and ultra-violet illumination in 

 various test runs. In general, he found that 

 there was no disadvantage and sometimes an 

 advantage to dropping some of the dial num- 

 bers. Fig. 10 compares the dial pairs foi- 



