78 



EXPERIMENT STATION. 



[Jan. 



fairh^ constant results are secured, according to our expecta- 

 tions. Tlie records at the extreme ends of tlie scale, especially at 

 the upper end, deviate considerably from the average, but this 

 was to have been expected, and for present purposes these rec- 

 ords may fairly be excluded. It seems proper further to throw 

 out three or four other records Avhich on account of excessive 

 deviation are open to suspicion. With these apparentl}^ abnor- 

 mal records temporarilj^ eliminated (they are to be studied 

 further in additional experiments), we secure results which 

 are rather remarkably uniform. Thus, the constant sum tem- 

 perature being computed to be 2,530 degree-hours, the extreme 

 deviation is less than 100, and the standard deviation is only 

 3(3.6°. The probable error is only 24. 7°.^ 



Thus, the temperature being known, and lying within reason- 

 able limits, the germination period of cress can be computed in 

 advance within a range of approximately two hours. 



The tabulated records used in these computations are ap- 

 pended to this report. 



A few abnormal records have been reserved for fm'ther study, 

 and there have arisen one or two intricate questions relating to 

 the whole theory of the physiological constant which must be 

 investigated further ; but as the figures stand they seem to 

 represent a considerable advance in this interesting field. 



Tabulation of Tliermograph Records. 



Elapsed Time (in Hours) for Germination 

 Phase of Cress. 



atcumulated temperature, in 

 Degree-hours. 



Above 38.2°. 



87, 



90, 



97, 



97, 



98, 



99, 



100, 



102, 



103, 



'2,.".1C 

 2,590 

 2,533 

 2,.562 

 2,518 

 2,.i72 

 2,.")07 

 2,.589 

 2,535 



1 "Standard deviation" is the geometric mean of deviations. " Proljable error" is de- 

 fined as tliat (l('i)arturc from tlie mean, on eitlier side, wilbiu wbich exactly oue-lialf tlie 

 variates arc found. 



