We cannot at the present point determine rigorously and in detail what is meant by an emotion (see Part I., Chap. vii.); it is enough to give a rough but comprehensible definition. From our standpoint, _emotion is in the order of feeling the equivalent of perception in the intellectual order_, a complex synthetic state essentially made up of produced or arrested movements, of organic modifications (in circulation, respiration, etc.), of an agreeable or painful or mixed state of consciousness peculiar to each emotion. It is a phenomenon of sudden appearance and limited duration; it is always related to the preservation of the individual or the species—directly as regards primitive emotions, indirectly as regards derived emotions. Emotion then, even while we keep to its primitive forms, introduces us into a higher region of the affective life in which its manifestations become complex. But how can we determine these primitive forms—the simple irreducible emotions—for this is our principal aim? Many neglect this determination, or leave it to arbitrary chance. The old authors seem at this point to have followed a method of abstraction and generalisation which could only lead them to entities. It was an accredited doctrine among them that all the “passions” can finally be reduced to love and hate; we meet this thesis throughout. To reach this conclusion they seem to have brought together and compared the different passions, disengaged the resemblances, eliminated the differences, and by continued reductions abstracted from this multiplicity its most general characters.[6] ----- Footnote 6: Descartes is a brilliant exception to this method of procedure; later on we shall have to consider his method (Part II., Chapter vii.). ----- If by love and hate we are to understand the movements of attraction or repulsion which lie at the bottom of the emotions, there is nothing to be said; but we are only given abstractions and theoretical concepts; such a determination is illusory and without practical utility. If we understand love (what love? for nothing is vaguer than this word) and hate in a more concrete sense, and pretend to consider them as the primitive source from which to derive all the other emotions, that is a purely mental opinion, an assertion which nothing justifies. The determination of the primitive emotions must be made not by abstraction and generalisation, but by _verification_. To attain this I can see but one method to follow—the method of observation, which teaches us the order and the date of appearance of the various emotions, and gives us their genealogical and chronological list. We may count as primitive all those which cannot be reduced to previous manifestations, all those which appear as a new manifestation, and those alone; all the others are secondary and derived.