IMPROVEMENT OF THE POTATO. 395 



investigations. A common error, as Thorne has pointed out (97), 

 is in disregarding the type and typography of the soil. The soil 

 should be as uniform as possible, preferably of sedentary origin, 

 and previous to its experimental use, should have been fertilized 

 and cropped in exactly the same manner. The land should be 

 slightly rolling and the potato rows should run up and down the 

 declivity so that the comparison of the whole rows is perfectly fair. 

 Water is a very important factor in potato growing, and small 

 depressions in flat lands receive an unequal portion of the surface 

 drainage which vitiates the results. 



E. Pagnoul (77) attributed the large yield of potatoes in sea- 

 sons when the total hours of sunshine were large, to the effect of 

 light on elaborating starch. His results may be slightly distorted 

 as he reports only a portion of the conditions which enter into his 

 experimental error, but results as reported gave a ratio of approxi- 

 mately 1 15 :n for plants under darkened glass, clear glass and open 

 air respectively. Therefore precaution should be taken concerning 

 all shade. 



Results from numerous experiments, among the most accurate 

 of which are those of Maw (74) and Plumb (82), have shown that 

 much care must be taken to have equal weight of seed planted, for 

 other conditions being equal, the yield of tubers increases directly 

 with the weight of the seed piece although not exactly proportional. 



Arthur (2, 3) and others have pointed out a difference in yield 

 owing to the difference in age and vigor of eyes sprouted before 

 planting. Wilting tubers up to a loss of moisture of 20 percent also 

 increased growth. He advances the proposition that : "Whatever 

 increases rate of growth at the beginning, increases yield." Finally 

 Gilmore (44) has shown that depth of planting certainly affects the 

 quality and on certain soils the yield; while Clinton (16) suggests 

 that it may also cause 'a difference in time of infection with Phy- 

 tophthora infest ans. 



These factors can probably never all be controlled in the same 

 experiment but they should all be kept in mind, and all reasonable 

 effort made to lessen errors due to them in comparative tests. 



CORRELATIONS 



Many so-called correlations of characters have been observed 

 by different investigators. These correlations are said to be some- 

 times 'very marked, and of great value in making selection 

 where one of the correlated characters is easier to select from 

 than the other. The physiological reasons for many of these ef- 



