CONDITIONS OF AN EXPERIMENT. 399 



for this proof, and the truly scientific investigator is ill-content 

 with theorizing not so supported. When the experiment has 

 this object, it will not be satisfied unless every step in the 

 reasoning is so supported. Reasoning upon premises afforded 

 by analogy is much open to error, and is a slovenly course, 

 when facts in the line of inquiry are to be had. For instance, 

 suppose one asserts that in-breeding in the horse is desirable, 

 is it any denial to show that in-breeding of swine is bad? 

 Certainly not, unless it is first shown that the two classes of 

 animals are so related, that Avhat afifects one for good or evil 

 will affect the other in like manner. Experiments should 

 follow a line of inquiry closely and evenly, and not be put off 

 the track by analogous reasoning, without the fullest research 

 to discover whether similarity holds in the comparison, — if 

 likeness exist only seemingly, and not in reality. Analogy, 

 so readily taken hold of by the imagination, leads astray. 

 Experiments, to be convincing, require that they be subject to 

 well-defined, clearly perceived conditions. When experiments 

 cannot be .repeated with like results, it shows that like con- 

 ditions were not present, or that we are not knowing and 

 regardful of conditions that vary the results of repeated 

 trials. It is well worth the care of any one who should wish 

 to be sure, as every one should, of conferring a benefit upon 

 his fellows, rather than risk doing them an injury, to delay 

 the expression of results of experiments until the correctness 

 of their results are tested by repeated trials. We venture the 

 remark that the Massachusetts Board should not give currency 

 to experimental efforts not fortified in this manner. 



Escape inferences when possible. The investigator asks ; 

 nature responds. If he would know something more, let him 

 ask again ; but do not think to support your reasoning, here 

 with an experiment, there with an opinion. To allow a free 

 rein to fancy in parts of an experiment, though good reason- 

 ing prevail in other parts, will vitiate the result. It is not 

 demanded that the investigator shall suppress his imagination ; 

 this would be at the risk of valuable suggestions ; but it is 

 required of him that he have it under control, so that his 

 eff'orts of reasoning may not become confused and valueless 

 by the irruption of a lively fancy. It is the weakest links 

 that determine the strenjfth of a chain. 



