'crawl–sound of word,' 'wriggle–rig,' 'fang–hang.' Still another indication of an effort to avoid significant associations lies in repeating the previous reaction word. Thrice this betrayed the object seen: once when 'wriggle' and 'fang' both suggested 'lizard,' once when 'medicine' was suggested by both 'bottle' and 'cork,' and once when 'poison' and 'rattle' both suggested 'antidote.' In one series the observer took refuge in the word 'animal,' to conceal the fact that she had seen the snake; 'animal' was given in response to 'gray,' 'bite,' 'rattle,' and 'adder.' Once the stimulus word itself, 'hour,' was repeated, leading to the correct conclusion that the watch was the object seen. Finally, suspicion was aroused in one instance because the reaction word seemed to show a reversion back to the preceding stimulus word: 'tick' suggested 'watch' and the next word, 'second,' suggested 'ticking, but the conclusion that the observer had seen the watch proved to be untrue.
It is thus evident how uncertain are the inferences based on the character of the associations alone. Judging from these results, the most trustworthy evidence is furnished by a misunderstanding of the stimulus word in a way that would be practically impossible for a person with the object to which it referred in mind, as when 'write,' which referred to the ink bottle, was heard 'right' and suggested 'wrong;' and by the occurrence of associations based on the sound of the word. Yet if 'write' had suggested 'wrong' only after an abnormally long interval, one might correctly infer that this was a forced association, so that it would be necessary to take account of the reaction time here also; while it may have been mere chance that the rhyming associations furnished the basis of none but correct judgments in these experiments of ours. There seems to be no a priori reason why undue emphasis on the sound of a word should always mean a desire to avoid associations connected with its meaning.
The best single criterion on which to base a decision as to the object seen by the observer in these experiments was the longest single reaction time. If, in each series, the longest reaction time was selected, and was found to be that of the reaction to a relevant word, this word proved in every case but two out of the fifty-three to refer to the object seen by the observer. If the longest time belonged to the reaction to an irrelevant word, the next longest was taken, and so on until one that was connected with a relevant word was reached. The longest single reaction time to a relevant word proved, in every series but two, to be that of a reaction to a word referring to the object seen. This record is better than that of the average reaction times; it will be remembered that there were four exceptions to the rule that these are longest for words connected with the object looked at. The exceptions to the rule of the longest single reaction time were among these four. In one of the exceptional series, the object seen was the pack of cards; the longest reaction time was for the association 'earthquake—Messina,' earthquake being an irrelevant word. Next longest was that for 'slimy—ooze,' 'slimy' referring to the snake, which had not been seen. Both of these delays might be accounted for by the emotional significance of the stimulus words. In the case of the other exception, the object looked at was the watch, but the longest single reaction time was that for 'Carter's—people,' in which the stimulus word referred to the ink-bottle. Here the delay was doubtless owing to the fact that a person who had not seen the ink-bottle would be somewhat puzzled by the unusual stimulus word 'Carter's.' There were eight cases where the longest single reaction time was that of the association with an irrelevant word, but where the longest
