If space permitted it would be easy to offer additional considerations tending to show that after infancy and early youth new acquisitions are mainly made by combinations and recombinations of ideas already possessed, and not by new and independent records physiologically reimpressed on each occasion. RULES FOR MAKING CORRELATIONS. 1. Never make a correlation except in conformity to In., Ex., and Con. Carelessness here is fatal to success. 2. When the pupil reads a correlation of mine, he should indicate the relations between the words by writing in the figures 1, 2, or 3, and he should pursue the same course with his own correlations. 3. Ofttimes "extremes" are in different planes of thought, so occasionally three intermediates are necessary to cement them; two are often required; but after considerable practice in making correlations one usually suffices. 1. What is fatal to success in making correlations? 2. What do the figures 1, 2, and 3 indicate in Rule 2? 3. How many intermediates should there be? 4. A correlation is a _successive advance_, and an intermediate must not refer back to any except its _immediate_ antecedent, never to its second or third antecedent. A pupil wrote:--_Short steps_ ... stepson ... real son ... more a son ... _Morrison_. Here, "more a son" refers to the comparison between "real son" and "stepson," but the latter is the second antecedent so the correlation is a defective one. He might have said: _Short steps_ ... _stepson_ ... _Morrison_. 5. A word may be used twice but never three times. _Pen_ ... pensive ... gay ... nosegay ... _Nose_. Here "gay" is properly used twice, and after that, it is dropped and you can go on with the rest of the word, to wit, _nose_. 6. A compound phrase including a verb is rarely allowable, since the intermediates must be the simplest elements, either sensations or perceptions [relations among sensations] or abstractions [relations among relations], or one of these with either of the others, always exemplifying either In., Ex., or Con.