cause it helps to pass the time, and keeps our thoughts from--
from--other things," he ended rather lamely.

For a few moments they remained silent and cogitating. Do what they
would to distract their minds from dwelling upon that ghastly scene in
the torture-chamber, the picture was constantly intruding itself upon
their imaginations; nor could they forget the ominous words of Alvarez
when he instructed the guard to keep them safely, as he would "require
them in the future."

But, after some little time spent in this dreary form of reverie, Roger
started up once more.

"Come, Harry," said he, "it is of no use for us to give way to these
miserable forebodings; let us get back to the cipher again. It will
keep us from thinking; and, besides, we may not have another such
favourable opportunity in the future."

Harry did not reply, but dismissed his gloomy thoughts, though evidently
with an effort, and once again the two leant over their precious paper
and cudgelled their brains in the effort to find the proper translation.

"Now," resumed Roger, "it seems to me that we may possibly be on the
correct track after all with our last grouping of the figures; that is
to say, adding the first two and last two figures in each group to one
another, and getting a certain number. It looks to me more likely than
any of the other methods we tried."

"I confess that I am unable to see it," answered Harry. "We have
already ascertained that by that method we get, first of all, the
numbers 49, 37, and 38, and, as I remarked, we have no letters
corresponding to those numbers."

"No, I agree with you there," replied Roger. "But how are we to know
that Jose selected the number 1 for his letter A, 2 for the letter B,
and so on? It is not very likely that he would, as that method of
procedure would make the solving of the cipher a fairly easy matter, and
we should have translated it by now. It is much more likely that he
took some other number for his letter A, say 15, or 40, or any number
rather than 1."

Harry retorted: "Well, in that case we are just as far off the solution
as at first, for how can we possibly tell, except by experiment, what
numbers correspond to the right letters? And it would take us weeks to
discover it by that method."

"I agree with you that we certainly should be a very considerable time
in arriving at the solution if we tried to do it simply by experiment,"
said Roger. "But I do not propose to se