such as
Hamlet's over the skull of Yorick. On the other hand, highly emotional
scenes are usually in verse, as are romantic passages like the
conversation of Lorenzo and Jessica in the moonlight at Belmont, or the
dialogues of Fenton and Anne Page, which contrast with the realistic
prose of the rest of the _Merry Wives_ and also the artificial
pastoralism of Silvius and Ph[oe]be in _As You Like It_. Few absolute
rules can be laid down in the matter, but study of Shakespeare's
practice reveals an admirable tact in his choice of medium.

[5] The figures here given are based in columns 1, 2, 3, and 4 on the
calculations of Fleay; in 5, 6, and 7 on those of Koenig; and in 8 on
those of Ingram. (S) = Shakespeare's scenes.

[Page Heading: Metrical Tests]

The frequency of rhyme, as shown in the fourth column, has more relation
to date. While there is no very steady gradation, it is clear that in
his earlier plays he used rhyme freely, while at the close of his career
he had practically abandoned it. The large number of rhymes in _A
Midsummer-Night's Dream_ and _Romeo and Juliet_ is accounted for mainly
by the prevailing lyrical tone of a great part of these plays, while, on
the other hand, in _All's Well_ it probably points to survivals of an
earlier first form of this comedy. It ought to be noted that, in the
figures given here, the rhyming lines in the play scene in _Hamlet_, the
vision in _Cymbeline_, the masque in _The Tempest_, and the Prologue
and Epilogue of _Henry VIII_ are not reckoned.

More significant are the percentages in columns five, six, and seven.
Before 1598, feminine endings never reach twenty per cent of the total
number of pentameter lines; after that date they are practically always
above that number, and show a fairly steady increase to the thirty-five
per cent of _The Tempest_. The variations of run-on lines (which, of
course, carry with them the frequency of pauses within the line, and
inversely the growing rarity of end-stopped lines) are closely parallel
to those of the feminine endings; while the increase in the proportion
of speeches ending within the line is still more striking. In _The
Comedy of Errors_ this phenomenon hardly occurs at all; in _The Tempest_
it happens in over eighty-four per cent of the speeches, the increase
being especially regular after 1598. Yet in some cases other causes are
operative. Thus cuts and revisions of plays were apt to leave broken
lines at the ends of spee