Transpiration as a Factor in Crop Production. 33 
not to introduce the additional error of overcropping and acci- 
dental or unintentional lack of uniformity in cropping. 
Among the investigators who have made extensive tests, Briggs 
and Shantz have uniformly replicated their potometers in a given 
year more than other investigators, usually averaging together 
six similarly treated pots. The Nebraska potometers have all 
been replicated from four to eight times barring an occasional 
injured plant, with four as the standard number. Occasionally, 
potometers contain broken or diseased plants and should be 
discarded in compiling averages. 
STAGE OF MATURITY. 
This factor has been discussed as a source of error in connec- 
tion with methods on page 25 of this bulletin. It was concluded 
that results with seedling plants are not dependable. The total 
water requirement up to the normal time for harvesting as a 
crop is of greatest practical interest. This requires mature plants 
with grain crops. In potometers too small to grow the crop 
normally to maturity, it should be harvested at a stage before the 
amount of soil has become an abnormal limiting factor. 
Conclusive experiments have not been made to show the 
relative water requirement at different stages of maturity. Fest 
(1908), Von Seelhorst (1908, 1910), and Thorn (1913) have in- 
vestigated the effect on water requirement of harvesting at different 
stages, but have not discriminated as to what part of the difference 
in water requirement is due to climatic differences and what part 
to the stage of development. 
ERRORS DUE TO METHODS OF COMPUTATION. 
A rather common error in compiling data which results in 
incomparable summaries and wrong conclusions is the placing in 
an average column the averages for different crops or treatments 
which do not cover the same period of years. As an illustration, 
the following table has been compiled from King's reports. 
