In order to construct these formulæ we must return to the set of questions which we employed in grouping the facts, we must answer each question, and compare the answers. We shall then combine them into as condensed and as precise a formula as possible, taking care to keep a fixed sense for every word. This may appear to be a matter of style, but what we have in view here is not merely a principle of exposition, necessary for the sake of being intelligible to the reader, it is a precaution which the author ought to take on his own account. The facts of society are of an elusive nature, and for the purpose of seizing and expressing them, fixed and precise language is an indispensable instrument; no historian is complete without good language. It will be well to make the greatest possible use of concrete and descriptive terms: their meaning is always clear. It will be prudent to designate collective groups only by collective, not by abstract names (royalty, State, democracy, Reformation, Revolution), and to avoid personifying abstractions. We think we are simply using metaphors, and then we are carried away by the force of the words. Certainly abstract terms have something very seductive about them, they give a scientific appearance to a proposition. But it is only an appearance, behind which scholasticism is apt to be concealed; the word, having no concrete meaning, becomes a purely verbal notion (like the soporific virtue of which Molière speaks). As long as our notions on social phenomena have not been reduced to truly scientific formulæ, the most scientific course will be to express them in terms of every-day experience. In order to construct a formula, we should know beforehand what elements ought to enter into it. We must here make a distinction between general facts (habits and evolutions) and unique facts (events). III. General facts consist in actions which are often repeated, and are common to a number of men. We have to determine their _character_, _extent_, and _duration_. In order to formulate their character, we combine all the features which constitute a fact (habit, institution) and distinguish it from all others. We unite under the same formula all the individual cases which greatly resemble each other, by neglecting the individual differences.