It may be recalled that samples from four types of film were stored for 
periods up to twelve months, at temperature levels of 70°, 90°, 120° F. The 
averages of differential shrinkage were compiled from the published graphs, and 
are given in Tables 4, 5 and 6, in arbitrary units. 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
Table 4 
(Temperature 70° F) 
| | 
3 Months | 6 Months 9 Months 12 Months Totals 
| 
AB-244 | 9.5 | 10.0 10.0 | 10.0 | 35.5 
AB2S |. 509 |. 45 4.5 Sov] 190 
AP-311 | 11.5 | 7.5 7.5 7.5 | 34.0 
Nos | 4 | 35 7.5 35° | 165 
Totals 29.0 |. 255 25.5 27.0 | 107.0 
Table 5 
(Temperature 90? F) 
3 Months | 6 Months 9 Months | 12 Months Totals 
| | 
AB-244 9,5 (0.0 | 10.0 9,5 39.0 
AB-284 3.5 5.5 | 5.0 5.5 19.5 
AP-311 3.5 3.5 | 3.5 7.5 18.0 
N-045 75 8 dv 180 7.0 36.0 
= | zZ a |- E N 
Totals 24.0 | ad: 225 |...,365 29.5 112.5 
Table 6 
(Temperature 120? F) 
3 Months 6 Months | 9 Months | 12 Months Total 
AB-244 | 20.0 23.5 23.5 24.5 91.5 
AB-284 | 3.5 3.5 | 3.0 5.0 15.0 
AP-311 | 7.5 7.5 | 11.0 3.5 29.5 
N-045 | 14.5 18.0 | 25.5 14.5 72.5 
Totals. 45.5 82:5 | 63.0 47.5 208.5 
| | 
  
  
  
  
The data are suitable for further analysis by applying the method of ‘ana- 
lysis of variance’. We shall first treat each Table individually, assuming the absence 
of the rest of the data, i.e. as if the experiment were carried out at only one tem- 
perature level. We shall later assess the gain due to carrying out the experiment at 
the three levels. 
The principle of the analysis is quite simple: The sum of the squares of the 
deviations of the tabulated magnitudes from their general mean can be decomposed 
14