184 CARNEGIE INSTITUTION OF WASHINGTON. 



\Lcttcr of Karl Pearson.^ 



University College, I^ondon, England, 



June 24, igo4. 



Dear Sir : I have put together a few suggestions that occur to 

 me, principally based on my own personal experience ; but I do not 

 wish them to be considered in any way as dogmatic statements, only 

 as impressions. 



(i) I agree absolutely with Professor Newcomb's first statement 

 that the nineteenth century has industriously piled together a vast 

 mass of astronomical, physical, and biological data, and that very 

 little use has hitherto been made of this material. The reason for 

 this I take to be that a man of mediocre ability can observe and col- 

 lect facts, but that it takes the exceptional man of great logical power 

 and control of method to draw legitimate conclusions from them. 



(2) Differing probably from Professor Newcomb, I hold that at 

 least 50 per cent of the observations made and the data collected are 

 worthless, and no man, however able, could deduce any result from 

 them at all. In engineers' language we need to "scrap" about 50 

 per cent of the products of nineteenth century science. The scientific 

 journals teem with papers which are of no real value at all. They 

 record observations which can not be made of service by any one, 

 however able, because they have not been undertaken with a due 

 regard to the safeguards which a man takes who makes observations 

 with the view of testing a theory of his own. In other cases the 

 collector or observer is hopelessly ignorant of the conditions under 

 which alone accurate work can be done. He "piles up" observa- 

 tions and data because he sees other men doing it and because that 

 is supposed to be scientific research. 



(3) I have had to deal to a great extent with the observations 

 and data of other men in my statistical laboratory, to which appli- 

 cations are always being made for aid in the interpretation of obser- 

 vations. I think I might help to illustrate my point by citing a few 

 actual experiences. 



(a) Meteorological Statistics. — We have here a large work in prog- 

 ress. The data are enormous, but without any .system. Examina- 

 tion shows that in Europe and America the returns are often un- 

 trustworthy. There is no standardization of method, of time, or of 

 quantity observed. Important stations are omitted or dropped for 

 years, and where a well-organized plan for a quarter of the expense 

 and labor would have led to definite results, the existing chaotic 



