Purely Literary rather than historical evidence should not be excluded if it helps to solve a problem, or even if it merely sharpens the sense of such a problem as the meaning of the dragon in Beowulf.
Some difficulty has probably been caused by thinking of the dragon as a symbol rather than as some other kind of image although, to be sure, a hard-and-fast distinction may not always be possible. To me the meaning of a symbol is more or less arbitrary, especially because the symbol represents something, or means something, usually quite unlike itself–as the eagle is physically unlike the United States. This meaning is imposed outside the work of art rather than within it. Such meaning has to be agreed upon, either by authors like Blake or Yeats in contracts with themselves which must sooner or later be registered with the people, or else by the folk in some other manner, as when they agree in a ‘religious’ community that, say, Ceres shall represent corn and harvesting. Once agreed upon, the meaning of the symbol must remain fixed, at least in the community; otherwise, communication will be confused. Of course, the determination of a symbol's meaning can only be by external evidence. Such meaning probably depends upon remote associations which are certainly not immediately available to the reader and which are, possibly, not always exactly available even to the writer, as in surrealism. These associations can often be recovered, no doubt, and the reason for the meaning can then be understood. But at least until one examines the creative processes behind the evolution of the symbol, its meaning must seem artificial.