SOIL RECLAMATION AND IMPROVEMENT 99 



to reclaim many types of land on an industrial system is 

 very often unpromising, as the capital necessary to be sunk 

 is too large in proportion to the returns. The whole question 

 of reclaiming land is of little value without considering some 

 system of experiment. The mere fact that a piece of land 

 is not doing well suggests the idea that probably somebody 

 has failed to do better, and that the case is, therefore, not 

 a simple one ; but it need not necessarily be very compli- 

 cated, and a simple type of experiment will not infrequently 

 solve the riddle as to its failure. A soil is so variable that if 

 any knowledge is required within a reasonable time, it is 

 necessary to conduct all experiments in duplicate. It is 

 not necessary that the piece of land under experiment 

 should be very large. The most important consideration 

 is that the person responsible for the experiment should have 

 a sound knowledge of the process of conducting experiments, 

 and a clear idea of the errors of practical experiment under 

 practical conditions. The chief type of such experiment 

 would be to lay out plots, of which there were two or three 

 plots containing no manure and no improving treatment 

 at all, two or three with phosphates, two or three with potash, 

 two or three with nitrogen, and two or three with lime. 

 Previous experience of that type of land would avoid many 

 unnecessary experiments, since at least some things might 

 be assumed fairly well beforehand, but soils differ so much, 

 and the causes of fertility and infertility are so many, that it 

 does not do to assume too much from a text-book. Roughly 

 speaking, the errors of experiment on a growing crop on a piece 

 of land will be about 10 per cent, of the yield. For the purpose 

 of the reclamation of barren land this is not at all a serious 

 error, since unless the land is going to double its capacity 

 it is hardly likely to pay any very heavy returns for big 

 initial expense. The difficulties are, therefore, not as great 

 as they are on an experimental farm. As the reclamation 

 of land is generally a matter of a fairly big scale, it would 

 be especially foolish to neglect a few preliminary experiments 

 before proceeding to effect some system of reclamation. 

 It is, of course, highly desirable that the materials used for 



