First of all we have the definite, assured, and certain fact that the earliest known instance of a crest is in 1198, and we find evidence of the use of arms before that date.
The next fact is that we find infinitely more variation in the crests used by given families than in the arms, and that whilst the variations in the arms are as a rule trivial, and not affecting the general design of the shield, the changes in the crest are frequently radical, the crest borne by a family at one period having no earthly relation to that borne by the same family at another.
Again, we find that though the occasional use of a crest can (by isolated instances) be taken back, as already stated, to a fairly early period, the use of crests did not become general until very much later.
Another fact is that, except perhaps in the persons of sovereigns, there is no official instance, nor any other authentic instance of importance, in which a crest appears ever to have been used by a woman until these recent and unfortunate days when unofficial examples can be found of the wildest ignorance of all armorial rules.
The foregoing may be taken as general principles which no authentic instance known can be said to refute.
Bearing these in mind, let us now see what other results can be obtained by deduction from specific instances.
The earliest form in which anything can be found in the nature of a crest is the lion upon the head-dress of Geoffrey, Count of Anjou (Fig. 28). This has been already referred to.
The helmet of Philippe D'Alsace, Count of Flanders (c. 1181), has painted upon the side the same figure of a lion which appears upon his shield.
What is usually accepted as the earliest authenticated instance of a regular crest is that afforded by the Great Seal of King Richard I. of England, which shows over the helmet a lion passant painted upon the fan-shaped ornament which surmounts the helmet.