Nor is it true that the Greeks made no use of experiment. The rise of the experimental method dates from the time when the medical schools began to influence the development of philosophy, and accordingly we find that the first recorded experiment of a modern type is that of Empedokles with the _klepsydra_. We have his own account of this (fr. 100), and we can see how it brought him to the verge of anticipating both Harvey and Torricelli. It is once more inconceivable that an inquisitive people should have applied the experimental method in a single case without extending it to the elucidation of other problems. Of course the great difficulty for us is the geocentric hypothesis from which science inevitably started, though only to outgrow it in a surprisingly short time. So long as the earth is supposed to be in the centre of the world, meteorology, in the later sense of the word, is necessarily identified with astronomy. It is difficult for us to feel at home in this point of view, and indeed we have no suitable word to express what the Greeks at first called an οὐρανός. It will be convenient to use the word “world” for it; but then we must remember that it does not refer solely, or even chiefly, to the earth. The later word κόσμος bears witness to the growth of scientific ideas. It meant at first the marshalling of an army, and next the ordered constitution of a state. It was transferred from this to the world because in early days the regularity and constancy of human life was far more clearly seen than the uniformity of nature. Man lived in a charmed circle of law and custom, but the world around him still seemed lawless. That, too, is why, when the regular course of nature was first realised, no better word for it could be found than δίκη. It is the same metaphor which still lives on in the expression “natural law.”[45] Footnote 45: The term κόσμος seems to be Pythagorean in this sense. It was not familiar even at the beginning of the fourth century. Xenophon speaks of “what the sophists call the κόσμος” (Mem. i. 11). For δίκη, see below, §§ 14, 72.