HOW TO CONDUCT AN EXPERIMENT. 311 



tion of sulphate of copper in very tall vertical glass tubes, 

 which are calculated to require several hundred years to 

 complete them. A multitude of experiments in the sub- 

 jects of crystallisation and liquid diffusion might easily 

 be devised and commenced, which would require thousands 

 of years for completion. 



Before commencing an experiment, a plan should be 

 formed of the mode of conducting it, i.e. of the order in 

 which the various changes should be made, and of the 

 notes to be taken. A memorandum should also be made 

 of the different substances or apparatus employed, their 

 conditions, forms, weights, sizes, positions, relations to 

 each other, temperature, or other circumstances which 

 may appear essential, in order that their effects may be 

 studied, and that the experiment may be exactly repeated, 

 if necessary, at any future time. In complex experiments, 

 or in those with dangerous substances, a rehearsal is often- 

 times necessary before the actual experiment, in order 

 that every attention may be paid to the more critical 

 points in the actual trial. I frequently adopted this plan 

 whilst investigating the properties of the extremely dan- 

 gerous substance, anhydrous hydrofluoric acid. Having 

 made all the necessary preparations, a preliminary trial is 

 then made ; and, having thus found that the apparatus 

 and materials will act, one of the first points to be deter- 

 mined is whether or not any of the obvious or well-known 

 causes of interference are influencing it. If, for example, 

 a galvanometer is used, and its needle moves, that instru- 

 ment should be disconnected, and the experiment repeated 

 without it. If the needle now moves, an interference 

 exists, and the galvanometer must be removed to such a 

 distance (sometimes as much as 30 or 40 feet) from the 

 experimental apparatus that an experiment with the latter 

 no longer affects it, except when the two are connected 



