before dinner, which he always had at night.
He kept a piano in the music room as well as at the log cabin; so if
he felt like working in the evening he could do so; and when he was
especially engrossed he often worked into the small hours. His
unselfishness made it easy for his wife, when she deemed a change and
rest essential, to make the excuse that _she_ needed it. After a
preliminary protest he would usually give in, and they would leave
Peterboro for a few days' excursion.

He knew discouragement in an extreme form. Many weeks, even months,
had to pass before his discontent over the last child of his
imagination would become normal. Particularly was this so with the
larger works; though each one was started in a fever of inspiration, a
longing to reduce to actual form the impossible. He was always
disheartened when a work was finished, but he was too sane in his
judgment not to have moments when he could estimate fairly the quality
of what he had written. But those were rare moments; as a rule, it was
in his future music that he was always going to do his "really good
work," and he longed ardently for leisure and freedom from care, so
that, as he once bitterly said, he would not have to press into a
small piano piece material enough to make a movement of a symphony.

His preferences in the matter of his own music were not very definite.
In 1903, when he had finished all that he was to write, he expressed a
preference for the "Dirge" from the "Indian" suite above anything that
he had composed. "Of all my music," he confessed at this time, "the
'Dirge' in the 'Indian' suite pleases me most. It affects me deeply
and did when I was writing it. In it an Indian woman laments the death
of her son; but to me, as I wrote it, it seemed to express a
world-sorrow rather than a particularised grief." His estimate of the
value of the music has, naturally, no extraordinary importance; but my
conviction is that, in this instance, his judgment was correct. As to
the sonatas, he cared most for the "Keltic"; after that, for the
"Eroica," as a whole; though I doubt whether there was anything in the
two that he cared for quite as he did for the Largo in the "Tragica"
and certain parts of the "Norse." He felt concerning the "Keltic" that
there was hardly a bar in it that he wanted changed, that he had
scarcely ever written any thing so rounded, so complete, in which the
joining was so invisible. He played it _con amore_, and it grew t