24 



THE SPECIFIC HEAT OF WOOD. 



given the ordinate indicated by its magnitude. Connecting these 

 in order gives an irregular broken line. But a regular curve, 

 whose form approaches that of the cubical parabola (a 2 y=x 3 ), 

 can be so drawn as to pass through many of the individual results 

 and very close to the remainder. The abscissa intercepted by 

 this curve at any ordinate indicates the corrected number of results 

 below the value of the ordinate taken. The corrected number of 

 results in each class (ABC, BCD, CDE, etc.) can thus be obtained 

 from the curve. This distribution differs from the observed by the 

 amount which the curve departs from the points plotted. The cor- 





IT 



*** 



o 





2 



7S o Off 90 95 



*o as So AS -*o 45 Jo 

 OP 



60 6S 



FIG. 4. Curve for correcting distribution of results for a smooth frequency curve. 



rected distribution, read from this curve, is nearer the truth than the 

 observed distribution because the irregularities in the observed dis- 

 tribution are unnatural and would disappear if the number of points 

 were greater. 



The frequency curve shown in figures 3 and 5 is obtained from the 

 corrected distribution of results derived from figure 4. It is entered 

 on figure 3 merely to show its relation to the broken lines indicating 

 the distribution of the results actually observed and is repeated on 

 figure 5 for further discussion. This curve furnishes the best infor- 

 mation which can be secured regarding the reliability of the mean 

 value of the specific heat of wood. 



