METHODS FOE PROVING ORIGIN OF SPECIES. 27 



These, if examined with care, may be very important. 

 The assertions contained in the works of historians or 

 even of naturalists respecting an epoch at which science 

 Avas only beginning, have not the same value. Lastly, 

 the common names are only an accessory means, especially 

 in modern languages, and a means which, as we have 

 seen, is not entirely trustworthy. So mnch may be 

 said in a general way, but in each particular case one 

 method or the other may be more or less important. 



Each can only lead to probabilities, since we are 

 dealing with facts of ancient date which are beyond 

 the reach of direct and actual observation. Fortunately, 

 if the same probability is attained in three or four 

 different ways, we approach very near to certainty. The 

 same rule holds good for researches into the history of 

 plants as for researches into the history of nations. A 

 good author consults historians who have spoken of 

 events, the archives in which unpublished documents are 

 found, the inscriptions on ancient monuments, the news- 

 papers, private letters, finally memoirs and even tradition. 

 He gathers probabilities from every source, and then 

 compares these probabilities, weighs and discusses them 

 before deciding. It is a labour of the mind which requires 

 intelligence and judgment. This labour differs widely 

 from observation employed in natural history, and from 

 pure reason which is proper to the exact sciences. 

 Nevertheless, when, by several methods, we reach the 

 same probability, I repeat that the latter is very nearly 

 a certainty. We may even say that it is as much a 

 certainty as historical science can pretend to attain. 



I have the proof of this when I compare my present 

 work with that which I composed by the same methods 

 in 1855. For the species which I then studied, I have 

 now moi'e authorities and better authenticated facts, 

 but my conclusions on the origin of each species have 

 scarcely altered. As they were already based on a 

 combination of methods, probabilities have usually 

 become certainties, and I have not been led to conclusions 

 absolutely contrary to those previously formed. 



Ai'chseological, philological, and botanical data become 



