between Stars forming Binary or Multiple Groups. 403 



Again, he calculates that the odds against any (two ?) such stars 

 fortuitously scattered falling within 32" of a third, so as to 

 constitute a triple star, is not less than 173,524 to 1. Now, 

 four such combinations occur in the heavens." Sir John 

 Herschel, from whose Outlines of Astronomy I take this state- 

 ment of Struve' s results, adds, "the conclusion of a physical 

 connexion of some kind or other is therefore unavoidable*." 



6. Now I desire it to be most distinctly understood, that 

 the argument which I have to state is not meant to controvert 

 the truth of the general result at which Mitchell and Struve 

 arrive, namely, that the proximity of many stars to one spot, 

 or the occurrence of many close binary stars distributed over 

 the heavens, raises a probability, or rather we would call it an 

 inductive argument, feeble perhaps, but still real, that such 

 proximity may be actual, not merely apparent; but I deny 

 that such probable argument is capable of being expressed nu- 

 merically at all; and I hope clearly to prove, that it has no 

 absolute and compulsory form addressing itself alike to all un- 

 derstandings and to all capacities, and to persons ill and well- 

 informed alike; but that the grouping of stars is like any 

 phenomenon occurring in physical investigations, which sug- 

 gests further inquiry ; which points at a result not improbable, 

 but requiring to be inductively established by bringing together 

 other considerations, whose accumulation may impel convic- 

 tion upon minds capableof estimating logical evidence, although 

 this evidence is no more capable of being expressed by num- 

 bers than is the evidence for the truth of the theory of gravi- 

 tation. 



7. These convictions were stated in a short letter addressed 

 (principally, however, in the language of doubt and inquiry) 

 to the Editors of the Philosophical Magazine and Journal, 

 which was published in that periodical for August 1849. I 

 had not then read Mitchell's paper, nor had I studied his ar- 

 gument; my incredulity was based on the general logical 

 ground that the data, as stated, do not contain the elements 

 for arriving at a numerical estimate of the probabilities for or 

 against such results. Having since considered the matter at 



* Outlinesof Astronomy, p. 564. 8vo. Longmans, 1849. I have thought 

 it quite necessary to recite this passage, which includes the reasoning, or 

 results of reasoning, to which I object; because I have been supposed to 

 refuse my assent to reasonings by no means of the same class, or liable to 

 the same objections as those contained in these sentences, which were 

 specifically quoted in my letter to the Editors of the Philosophical Maga- 

 zine mentioned in art. 7- M. Struve and Sir John Herschel have only 

 adopted and extended the conclusions implicitly received by many illus- 

 trious predecessors, and are of course not more individually answerable for 

 them than they. 



2 D 2 



