In 1936 he wrote The Study of the History of Science, a version of inaugural lectures he gave at Harvard. Some of the contents are as fresh today as they were when first articulated:

...we shall not be able to understand our own science of to-day (I do not say to use it, but to understand it) if we do not succeed in penetrating its genesis and evolution. Knowledge is not something dead and static, but something fluid, alive, and moving. The latest results are like the new fruits of a tree; the fruits serve our immediate practical purposes, but for all that it is the tree that matters. The scientist of philosophic mind is not interested so much in the latest results of science as he is in its eternal tendencies, in the living and exhuberant and immortal tree. The fruits of to-day may be tempting enough, but they are not more precious to his way of thinking than those of yesterday or to-morrow. (5)

Here is a fact. on March 13, 1781, William Herschel discovered a "curious either nebulous star or perhaps a comet: which turned out to be a new planet, Uranus. What does that mean? In order to appreciate it, one must bear in mind the histroy of astronomy down to that time and realize that this was the first planet to be discovered within the memory of mankind! One must bear in mind the circumstances of Herschel's life and activity. How did he discover Uranus, and why did nobody else discover it before him? What were the implications, for him and others, for us, of that discovery? THe statement of the discovery is almost meaningless, unless all these explanations are implicitly available to the reader, or explicitly stated to him. A bare fact is nothing, though its potentialities may be endless; meaningless to one person, it opens up to another a new chapter of science or life, and provides the kernel of a tragedy. the historico-scientific facts are the stones of the building which the historian of science is building; if accurate, they are invaluable --history without them is nothing but a vision, a dream, a bubble, a shadow, and even less, the shadow of a shadow; yet they are worthless unless placed in their proper sequence and position, each one being explained and justified by all the others. (32-33)

...let us consider for a moment the idea of experimentum crucis. One calls a crucial experiment one which enables him to choose between hypotheses which are mutually exclusive by proving that one of them is right and that the other must be wrong. The classical example is that concerning the wave theory of light and the emission theory. There was much discussion in the first half of the nineteenth century as to which of those theories was right to the exclusion of the other. The latter had been favored by Newton and was generally accepted by the triumphant Newtonians; the first had been brilliantly but incompletely explained by Huygens, and after a century of neglect it had been revived by Young and almost completely vindicated by Fresnel. A few years after the latter's death, Sir William Rowan Hamilton in the course of his analytical development of the wave theory was able to predict the existence of a very rare kind of refraction, the conical refraction, which had never been observed. Hamilton's mathematical prediction of 1832 was verified experimentally by Humphrey Lloyd in the following year. The wave theory seemed to be established on an inexpugnable basis, and yet some of the defenders of the emission theory refused to capitulate. It was then that Arago invented an ingenious experimentum crucis. If the emission theory is correct, the speed of light must increase with the density of the medium; if the wave theory is correct, the speed must decrease with the density. Hence if we could measure the speed of light in air and water, the comparison of the results would tell us which theory was true. The extremely delicate experiments which he suggested were not realized until fifteen years later by Foucault, who proved that the speed of light is smaller in water than in air, and thus "that the emission theory is incompatible with the reality of the facts"... This seemed to be decisive and final. there is no doubt that the emission theory was incompatible with the facts which he dealt with, but it did not follow that that the wave theory was compatible with every other fact. To make a long story short, the study of back-body radiation revealed facts which were incompatible with the wave theory, and led to the formulation of the quantum theory by Max Planck in 1900. Arago's experimentum crucis had simply proved that the wave theory was more complete than the emission theory, but it had not proved and could not prove that it was absolutely true. (38-39)