s, unless, of course, Alvarez should send for us; but
I do not think he will want to question us to-day; he has not yet
finished with that poor wretch de Soto. Now, Harry, just rip up the
seam of my jerkin, and get that paper out, and let us start the business
at once."

Harry took out his knife, which, strangely enough, he had been permitted
to retain, and, carefully cutting the stitches, removed the paper,
unfolded it, and laid it open upon the stone table. Then both lads
leant over the document and concentrated all their energies on the task
before them.

"First of all," said Roger, "what language is it most likely to be
written in? Jose Leirya was himself a Spaniard, it is true; but from
what I could gather from that man William Evans, about whom I told you,
his crew was invariably made up for the most part of Englishmen; so it
is reasonable to assume that English would be the language he would have
to employ on board his ship. He had been sailing the high seas as a
pirate for a good many years; so one would imagine that at the time when
he wrote that cipher he would know probably more English than Spanish.
What, therefore, more natural than that he should write his secret in
English? At any rate, I think we should not be far wrong in assuming
that it is written in English; and so we will take it for granted that
such is the case. And if we find that we are wrong, we will try some
other language--say Spanish.

"But the language is not all-important; it is the finding what signs or
letters those figures stand for that will be the difficulty. Now let us
have a look at the paper. There is the first line of figures.

"1581. 2227 1819 1919 2622 1820 1335 1138 1918.

"Let us take that first, and see what we can make of it. I should say
that the first number, standing, as it is, by itself, is the year in
which it was written, that is to say, the year 1581. Now, you observe
that these figures are all in groups of four. We will say that each
figure represents a letter, which is not very likely, as not all the
words could possibly consist of four letters each; but they might be the
initial letters of certain words, giving sufficient of the word to
enable one to guess the rest. Now there are 26 letters in the alphabet.
Taking A as being 1, B as 2, C as 3, and so on up to Z as 26, let us
apply this to the cipher.

"By doing this with the first group, we get B B B G, or, if we take the
figures in groups of two