Thünen distinguished between insurable and 
uninsurable risk, so the basic idea was known in 
1850; Thünen writes, "there exists no insurance 
company that will cover all and every risk 
connected with a business. A part of the risk 
must always be accepted by the entrepreneur" 
(1960 edition, p. 246). Knight refers to Thünen 
in Part I, Chapter II of Risk, Uncertainty, and 
Profit, as well as the economists A. H. Willett 
and A. S. Johnson, both (according to Knight) 
students of J. B. Clark. Knight characterizes 
their books (Willett, Economic Theory of Risk and 
Insurance, 1901; Johnson, Rent in Modern Economic 
Theory, 1902) as on the right track, but 
incomplete; while they recognize the distinction 
between insurable and uninsurable risks, they 
treat the latter as "known" quantities.

The frequentist approach to probability was also 
emerging around the time of Knight's treatise. 
Richard von Mises's distinction between "class 
probability" and "case probability" appears most 
clearly in his Probability, Statistics, and 
Truth, published in 1939, but the basic idea (I 
think) appeared in a 1919 article by Mises. 
Keynes's Treatise on Probability, published the 
same year as Knight's book, also has a chapter on 
frequentism, which might have some useful references.

Peter Klein