April 6, 18So.] 



SCIENCE. 



251 



Unfortunately the season of 1881 was ex- 

 tremely- dry. and the manures applied produced 

 scarcely any appreciable effect ; so that, al- 

 though various minor results of interest and 

 value were obtained, the main object of the 

 experiments was scarceh' at all advanced bj- 

 the year's work. The most interesting of these 

 minor results is, perhaps, the striking and bene- 

 ficial effect exercised on tlie yield of some of 

 the plots by the thorough drainage which they 

 received. Barnyard manure was the onlj- fertil- 

 izer which produced anj' noticeable effect ; and 

 this is ascribed rather to its physical action in 

 making the soil more retentive of water than 

 to anj' direct fertilizing action. 



It is evident that circumstances have cou- 

 S2:)ired to render this simplj' a preliminary re- 

 port, whose value consists in its account of the 

 plan and methods of the experiments more than 

 in any results yet attained. 



Dr. Miles appears to be fully aware of the 

 complex nature of the problems attacked, and 

 to have taken great care to execute all the 

 •operations of tillage, planting, cultivation, and 

 harvesting in a uniform manner on the several 

 plots. He is cautious, too, in drawing con- 

 clusions, and not in haste to attribute small 

 ■difference of yield to the effects of different 

 fertilizers, as is too often the case. 



His method of comparing the yields of a 

 manured and an unmanured plot is novel and 

 interesting. Instead of assuming the differ- 

 ence between the two to represent the effect of 

 the manures, as is usually done, he first grows 

 a crop on all the plots without manure. In 

 the crop of the succeeding year, he first notes 

 the gain or loss of j'ield on the unmanured 

 plot, and then assumes, that, if the plot to be 

 compared had not been manured, its yield 

 would have varied to the same extent. Then 

 the difference between the actual yield of the 

 plot and what it would have yielded without 

 manure is regarded as the effect of the fertiliz- 

 ers applied to it. The following example il- 

 lustrates the method : — 



This method of comparison is evidentlj' in- 

 tended to take account of the natural uneven- 

 ness of the soil, and it is to a certain extent 

 an improvement over the direct comparison of 



3"ields ; but it also involves errors of its own, 

 and not only that, but errors of unknown 

 amount. Because plot 3 jielded one bushel 

 per acre more than plot 1 in 1880, it is bj' no 

 means certain, that, in the veiy different season 

 of 1881, the same difference would have been 

 observed: indeed, it is highly probable that 

 it would not have been. Dr. Miles recognizes 

 this, and designates the 7.3 bushels of our 

 table as ' probable increase produced by ma- 

 nures.' But he gives us no means of knowing 

 whether this amount is within or without the 

 limits of error ; that is, whether the manure 

 on plot 1 actually did produce an effect or not. 

 This cannot but be regarded as a serious defi- 

 ciency in these otherwise valuable experiments ; 

 and it is one that no care in the execution 

 of the experiments can do anj^ thing to re- 

 move. 



A field-experiment with fertilizers involves 

 one of two assumptions, — either that the 

 several plots have exactly the same crop-pro- 

 ducing iDower, or that the differences observed 

 in a preliminary unmanured crop are constant. 

 Neither of these assumptions is true. With 

 the greatest care in the selection of plots, verj' 

 considerable differences in both respects will 

 show themselves. Such being the case, the 

 scientific conduct of a field- experiment requires 

 that the amount of error involved in the above 

 assumptions shall be determined, to the end 

 that we may know whether the apparent differ- 

 ences in the effects of the fertilizers have any 

 real significance. This may be done by mul- 

 tiplying the number of plots which receive the 

 same treatment, and distributing them uniform- 

 Ij' over the experimental field ; the only limit 

 to the multiplication being that imposed by 

 practical considerations of the possibility of 

 treating a large number of plots. 



In this waj- it is possible to obtain, not onl}' 

 the average yield of a certain fraction of an 

 acre under particular treatment, but the amount 

 of variation from that average which maj' be 

 expected in individual cases. This method 

 calls for a multiplication of the manured, as 

 well as of the unmanured plots : it greatl^^ 

 increases the labor of conducting a field-ex- 

 periment ; but the results, once obtained, are 

 reasonably accurate, and we know how accu- 

 rate they are. 



This whole subject has recently been very 

 thoroughly discussed by Wagner ; and a peru- 

 sal of his papers ' cannot fail to be in the high- 

 est degree interesting and suggestive to all who 

 contemplate making field-experiments. 



