Something on Chronology
From our winnowing of the chaff of ages have emerged a very few hard won grains of apparent fact. First and foremost, the bulk, and presumably all, of these surviving sonnets was written from about 1593 to about 1599. This conclusion rests fundamentally on long accepted literary relationships, controlled by external facts. The fundamental principle of these literary relationships has been to discover the basic source of origin and thence to determine direction of evolution, just as in manuscript relations. Division into series has no bearing here, but is an independent and correlative by-product of our examination. The majority of these fundamental relationships and all the external facts have long been known and accepted. I have simply gathered them and put them together to discover what their genetic pattern would tell us. The facts, both internal and external, are sufficiently numerous and have stood the acid test of scholarly examination for a sufficient length of time to neutralize the combined personal equations of previous scholars and to minimize my own.
Perhaps it should here be emphasized that in our study we have not used parallels as such; there is always a base from which to determine direction of development. The base may be found to be inadequate, the direction of development as indeterminate; that is, some of our instances may be found to be only parallels and so useless in this analytical framework. But parallels as such have not been used.
There have, of course, been efforts in the past to use parallels. Ordinarily, these would need to be reduced to statistical level in order to be interpreted. Such a test is that of twice-used words. It is a simple matter now to test this idea statistically. The letters A and B in Bartlett's Concordance will be sufficient. So far as I can see, each of the plays represented is about equally yoked with every other play by twice-used words. I find no significant statistical incidence at all to indicate relative chronological position. If twice- used words yield no statistical significance, neither, it would appear, could twice-used phrases as such, as advocated by Beckwith. At