right. You come to the third act, and somehow it won't go at all. You
battle with it for weeks in vain; and then it suddenly occurs to you,
'Why, I see what's wrong! It's that confounded scene where the man finds
the will under the sofa! Out it must come!' You cut it out, and at once
all goes smooth again. But you have thrown overboard the great effect
that first tempted you."]

[Footnote 6: The manuscripts of Dumas _fils_ are said to contain, as a
rule, about four times as much matter as the printed play! (Parigot:
_Génie et Métier_, p. 243). This probably means, however, that he
preserved tentative and ultimately rejected scenes, which most
playwrights destroy as they go along.]

[Footnote 7: Lowell points out that this assertion of Heminge and
Condell merely shows them to have been unfamiliar with the simple
phenomenon known as a fair copy.]

[Footnote 8: Since writing this I have learnt that my conjecture is
correct, at any rate as regards some of M. Hervieu's plays.]

[Footnote 9: See Chapters XIII and XVI.]

[Footnote 10: This view is expressed with great emphasis by Dumas _fils_
in the preface to _La Princesse Georges_. "You should not begin your
work," he says, "until you have your concluding scene, movement and
speech clear in your mind. How can you tell what road you ought to take
until you know where you are going?" It is perhaps a more apparent than
real contradiction of this rule that, until _Iris_ was three parts
finished, Sir Arthur Pinero intended the play to end with the throttling
of Iris by Maldonado. The actual end is tantamount to a murder, though