Why wasn't it reasonable to presume that in the last instant of Florey's life, just before the attack was made, he had attempted to conceal the script. He had thrown it from him; his death-cry had aroused the household so that the murderer had no time to seek and procure it. Then from a hiding place, or even from among a group of the guests, he had seen me pick it up.

To work out that cryptogram, to read its hidden meaning was the first and the best thing I could do in the way to solve the mystery of Kastle Krags. Written originally on parchment, sixty or seventy years before, it doubtless referred and was in explanation of the secret of the old manor house--the legend of the treasure, supposedly hidden by Godfrey Jason in the long ago. I had just toyed with it before. Perhaps I had had little faith that it was of any real importance. But now, other avenues had failed, and I was resolved to know the truth if it was humanly possible to do so. I copied the script again, with great care:

Kastle Krags — Part 91

aned dqbo aqcd trkm fipj dqbo seho ohuy wvyn dljn dtht

Kastle Krags — Part 91

Then I began to make a systematic analysis. I noticed first that the second and the sixth words were identical, indicating--considering the brevity of the entire message--that it must represent a word of most frequent use. Of course the articles "a" and "the" occur most often in any English writing, yet I found it hard to believe that "dqbo" represented either. In the first place, in a message of that length it is reasonable to assume that all articles and words not absolutely necessary to the meaning had been omitted.

Kastle Krags — Part 91

Weeks that seemed years before Nealman had told me that, after careful study, he had been convinced that there was some truth in the legend of buried treasure. Was it not within the bounds of reason to assume that this cryptic message revealed the hiding place of the treasure? Working on this assumption, I made up an imaginary description of some hiding place, just to see what words occurred with the greatest frequency. I found at once that the word that would be most likely to be used twice in a description of that kind would be some measurement--either feet, yards, meters, rods, or something of the kind. If I could convince myself that "dqbo" represented some English measurement I might find the key and system of the code.