Peter Kropotkin did it in 1902 according to Wikipedia on the phrase "Survival of the fittest."

He wrote:

If certain heritable characters increase or decrease the chances of survival and reproduction of their bearers, then it follows mechanically (by definition of "heritable") that those characters that improve survival and reproduction will increase in frequency over generations. This is precisely what is called "evolution by natural selection". On the other hand, if the characters which lead to differential reproductive success are not heritable, then no meaningful evolution will occur, "survival of the fittest" or not: if improvement in reproductive success is caused by traits that are not heritable, then there is no reason why these traits should increase in frequency over generations. In other words, natural selection does not simply state that "survivors survive" or "reproducers reproduce"; rather, it states that "survivors survive, reproduce and therefore propagate any heritable characters which have affected their survival and reproductive success". This statement is not tautological: it hinges on the testable hypothesis that such fitness-impacting heritable variations actually exist (a hypothesis that has been amply confirmed.)[23]

Andrew Sutton

So, what prince Kropotkin told us was that, assuming that the traits are heritable, the traits which are favourable for reproduction will propagate in generations and those which are not will not. That is, reproducing reproduce. A priory true and not testable. The only non-trivial assumption concerns heritability. Is the latter what is called "natural selection theory"?

Albert George Orr

So, in order to avoid tautology you need to give an operational definition of "functional efficiency" independent of survival. Then the two could be compared in experiments and thereby test the statement that "functionally efficient" have higher chances of surviving (reproduction).

F=ma isn't a testable law of nature, it is an operational definition of force. The postulate can be formulated as follows

Postulate 2

For any isolated system of N particles with coordinates x_i defined relative to an inertial

reference frame there exists a function U(x^N), called potential energy, such that its derivative with respect to x_i is equal to m_i a_i,

where m_i is a constant associated with the particle i. and a_i is the second time derivative of x*i.

These equations are called equations of motion.