We pass on to experiments in which the ideas transmitted from the agent find other subterranean channels in the percipient's organism for their expression. Of all forms of intelligent automatism writing, next to speaking, is probably in an educated percipient the easiest, because in normal life the commonest. In the cases, therefore, recorded below the actual movements involved, though of a relatively simple kind, as being unaccustomed called possibly for the exercise of a degree of mental activity as high as would have been the case had writing been the vehicle of expression. In the preceding chapter it was recorded, in the experiments with numbers, that some of the answers were given through the movements of a table on which the percipient's hands rested (p. 73). A series of experiments of this nature was made by the writer in November and December 1873, with the assistance of a few friends, amongst whom were Mr. F. H. Colson, now Head Master of Plymouth College, and the Rev. W. E. Smith, of Corton, near Lowestoft. The following is a description of the methods adopted. Three or four of us would sit round a small centre-legged table, cane-bottomed chair, waste-paper basket, or metal tripod, with our hands resting on it. We found that in a few minutes the table (or other instrument) would tilt on one side, or move round and round, with considerable freedom. When these motions had once been fairly established, one or two of those present in the room would retire to a distance, keeping their backs to the table, and think of a letter of the alphabet. The table would move freely up and down, under the varying pressure of the hands laid on it, in a succession of small tilts. Those sitting at the table would count the tilts--one tilt standing for A, two for B, three for C, and so on. Excluding second trials, there were 70 experiments conducted under these conditions. The right letter was tilted in 27 cases, and in two others the next succeeding letter was given. On some occasions the proportion of successes was much higher; thus, on the 28th November, out of a total of 16 trials, 10 were correct. On the 1st December, on the other hand, 10 trials were made without any success. It was the rule throughout that the agents should stand with their backs to the table at some distance from it, and after the first few experiments we found, or thought we found, that the thought-transference succeeded best with a single agent. In order that the letter might not be guessed from the context, we generally took the initial or initial and final letters only of a word; in four cases only did the agent select as many as three consecutive letters of a word. If the letters had been arbitrarily chosen, the chances against the right letters being indicated would be 25 to 1. But as the letters actually selected were in most cases constituent parts of a word, generally the initial letter, and as in some cases two or three consecutive letters were