14 PLANT PHYSIOLOGY 



should be the constant care of the student. The more promi- 

 nent of these may be classified thus: 



1. Errors oj the Instrument. Some of these are so crass as 

 to be obvious, such as crudeness, shakiness or leakiness due 

 to poor mechanical construction (very apt to characterize make- 

 shift or even adapted apparatus), and presence of dirt, or of 

 chemicals unremoved from earlier experiments. More serious 

 are errors of standardization, due either to the carelessness of 

 the maker or his desire to sell the instrument cheaply (as promi- 

 nently shown in cheap thermometers); the remedy is to buy 

 good articles from makers with a reputation to sustain, and 

 also and especially, to test the standardization by comparison 

 with instruments of known accuracy. A still more serious, 

 because so insidious, phase of this error comes from spontane- 

 ous change in the standardization. To some extent thermom- 

 eters, especially when newly made, are liable to this. In another 

 way, small weights are very susceptible to alteration through 

 adherent grease, etc. Self-recording meteorological instruments, 

 such as thermographs and hygrographs, are very liable to work 

 themselves out of true, and should often be standardized by 

 comparison with instruments of known accuracy. Errors, of 

 obscure cause, may intrude in control experimenting from the 

 use of dishes or other articles differing merely in size, shape, or 

 color, and these should always be chosen exactly alike even 

 when there is no conceivable reason why differences of this 

 sort should affect the result. 



2. Errors of the Surroundings. These are connected largely 

 with meteorological changes, which are incessant. Thus, read- 

 ings made at different temperatures may be vitiated by the 

 expansion or contraction of metal, or even glass, parts of the 

 instrument, and especially of mercury where that is employed. 

 Readings made at different humidities may involve a serious 

 error from the hygroscopic lengthening or shortening of papers 

 or of threads, especially if the latter are so used that their altera- 

 tion may be magnified. This happens commonly with the cords 

 of auxanometers, where the error may become so great as actually 



