The experiment is one of the many forms of the "completion test," or "the combination method." As we have already noted, the power to combine more or less separate and isolated elements into a logical whole is one of the most essential features of intelligence. The ability to do so in a given case depends, in the first place, upon the number and logical quality of the associations which have previously been made with each of the given elements separately, and in the second place, upon the readiness with which these ideational stores yield up the particular associations necessary for weaving the given words into some kind of unity. The child must pass from what is given to what is not given but merely suggested. This requires a certain amount of invention. Scattered fragments must be conceived as the skeleton of a thought, and this skeleton, or partial skeleton, must be assembled and made whole. The task is analogous to that which confronts the palæontologist, who is able to reconstruct, with a high degree of certainty, the entire skeleton of an extinct animal from the evidence furnished by three or four fragments of bones. It is no wonder, therefore, that subjects whose ideational stores are scanty, and whose associations are based upon accidental rather than logical connections, find the test one of peculiar difficulty. Invention thrives in a different soil. Binet located this test in year X. Goddard and Kuhlmann assign it the same location, though their actual statistics agree closely with our own. Our procedure makes the test somewhat easier than that of Binet, who gave only one trial and used the somewhat more difficult words _Paris_, _river_, _fortune_. Others have generally followed the Binet procedure, merely substituting for Paris the name of a city better known to the subject. Binet's requirement of a written response also makes the test harder. Perhaps the greatest obstacle to uniformity in the use of the test comes from the difficulty of scoring, particularly in deciding whether the sentence contains enough absurdity to disqualify it, and whether it expresses three separate ideas or only two. It is hoped that the rather large variety of sample responses which we have given will reduce these difficulties to a minimum.