Of these tests, that of rhyme and that of feminine endings, discreetly employed, are of use in broadly distinguishing Shakespeare’s plays into two groups, earlier and later, and also in marking out the very latest dramas; and the feminine-ending test is of service in distinguishing Shakespeare’s part in Henry VIII. and the Two Noble Kinsmen.  But neither of these tests has any power to separate plays composed within a few years of one another.  There is significance in the fact that the Winter’s Tale, the Tempest, Henry VIII., contain hardly any rhymed five-foot lines; but none, probably, in the fact that Macbeth shows a higher percentage of such lines than King Lear, Othello, or Hamlet.  The percentages of feminine endings, again, in the four tragedies, are almost conclusive against their being early plays, and would tend to show that they were not among the latest; but the differences in their respective percentages, which would place them in the chronological order Hamlet, Macbeth, Othello, King Lear (Koenig), or Macbeth, Hamlet, Othello, King Lear (Hertzberg), are of scarcely any account.[283] Nearly all scholars, I think, would accept these statements.

The really useful tests, in regard to plays which admittedly are not widely separated, are three which concern the endings of speeches and lines.  It is practically certain that Shakespeare made his verse progressively less formal, by making the speeches end more and more often within a line and not at the close of it; by making the sense overflow more and more often from one line into another; and, at last, by sometimes placing at the end of a line a word on which scarcely any stress can be laid.  The corresponding tests may be called the Speech-ending test, the Overflow test, and the Light and Weak Ending test.

I. The Speech-ending test has been used by Koenig,[284] and I will first give some of his results.  But I regret to say that I am unable to discover certainly the rule he has gone by.  He omits speeches which are rhymed throughout, or which end with a rhymed couplet.  And he counts only speeches which are ‘mehrzeilig.’  I suppose this means that he counts any speech consisting of two lines or more, but omits not only one-line speeches, but speeches containing more than one line but less than two; but I am not sure.

In the plays admitted by everyone to be early the percentage of speeches ending with an incomplete line is quite small.  In the Comedy of Errors, for example, it is only 0.6.  It advances to 12.1 in King John, 18.3 in Henry V., and 21.6 in As You Like It.  It rises quickly soon after, and in no play written (according to general belief) after about 1600 or 1601 is it less than 30.  In the admittedly latest plays it rises much higher, the figures being as follows:—­Antony 77.5, Cor. 79, Temp. 84.5, Cym. 85, Win.  Tale 87.6, Henry VIII. (parts assigned to Shakespeare by Spedding) 89.  Going back, now, to the four tragedies, we find the following figures:  Othello 41.4, Hamlet 51.6, Lear 60.9, Macbeth 77.2.  These figures place Macbeth decidedly last, with a percentage practically equal to that of Antony, the first of the final group.