54 AN INTRODUCTION TO SCIENCE 



experienced. But how is he to tell what the result would 

 have been had the season been wet instead of dry, or dry 

 instead of wet ; had there been less reserve of nitrogen in 

 the soil or more phosphate? Or, again, how is he to tell 

 whether kainit is equally useful for light soils and heavy, for 

 gravels and marls and clay? It is not the experiments which 

 cost trouble but the control. Anyone can say try x, or y, or 

 z ; it is only the trained experimenter who can say whether 

 and how far the result is due to the use of x, or y, or z. 



If, on the map of a certain country we are citing an 

 observation recently brought to our notice the extent to 

 which cancer is prevalent is marked by shades of grey, the 

 "cancer spots" are sufficiently dark to attract anyone's 

 attention. Such a map having been made, coincident con- 

 ditions were sought for, and it was observed that within 

 a certain area wherever these foci of the disease occur a 

 particular kind of tree (we will not say what kind of tree, 

 lest someone unversed in scientific method wage a crusade 

 against it) is abundant and grows near the houses. Is 

 there any connection between this tree and cancer ? Long 

 before the life history of "rust" had been worked out, 

 farmers held a conviction it was regarded as a vulgar 

 prejudice that their wheat was affected with rust in fields 

 bounded by hedges in which the common barberry grew. 

 It has since been ascertained that the fungus which in one 

 stage of its existence affects wheat with rust is in another 

 stage the aecidium or cluster-cup fungus of barberry ; and 

 it has been found not only that rust occurs where there are 

 barberries, but that it does not occur to the same extent 

 where there are none. The illustration is not altogether 

 satisfactory, for rust occurs in generation after generation 

 of wheat-plants in Australia and India, where the barberry 

 is not a native plant ; but this fact does not disprove 

 Du Bary's assertion that in England its choice of host-plants 

 alternates between wheat and barberry. It shows, however, 

 either that rust can dispense for a long period with the 



