The means of the readjustments made with the various colours are shown in the following table .sx In each series , whether the subject began with a more saturated or a less saturated colour , he stopped before he reached the standard which he had been instructed to remember , in other words , when he proceeded from a more saturated colour , he considered that a colour which was actually more saturated than the original was equal to it , and conversely when he proceeded from a less saturated one .sx In only 16 readings out of 166 was the final reading equal to the standard or beyond it :sx 11 of these were given by two subjects .sx This is very remarkable , for many of the subjects complained that they had forgotten the original colour :sx it is by no means easy to identify a shade in a gradually changing series .sx It may be objected that , as the subjects were instructed to stop as soon as they reached the standard and were discouraged from making a to and fro movement , the means shown , e.g. 30.8 and 19.8 in the naphthol yellow series , merely mark the point of entry into that region in which lie a large number of slightly different shades all indistinguishable from one another and the standard , what is commonly called the region of the difference threshold .sx This , however , would not account for the difference between the means , since a movement of one degree on the scale in the neighbourhood of the standard produced a perceptible difference .sx When the standard was shown a second time , the subject could always tell whether his setting had been correct :sx frequently he would say " Is that the standard ?sx If it was , I stopped too soon .sx " As it .sx was advisable , when further observations were to be made , that the subject should not know the nature or extent of his error , questions of this kind were left unanswered and the subject came to think that the standard was varied from one observation to another .sx It will be noticed that the difference between the standard and each of the means is greatest in the naphthol yellow series :sx this is probably because there is less variation in brightness in this series than in any other , and so the subject has to judge more by changes of hue than by changes of hue and brightness .sx The 'difference is , however , one which is not to be stressed , for it is possible to increase the size of any threshold as measured by the scale by decreasing the concentration of the solution .sx The difference between the standard ( 25 ) and the higher mean is in each case greater than that between the standard and the lower mean .sx This is partly explained by Weber's Law :sx equal intervals on the scale correspond to equal increases in the depth of the fluid seen through the prism , and the perceptibility of a difference in coloration depends , not on its absolute value , but on the ratio of the difference between the stimuli to one of them ; hence an increase of one on fifteen is perceived more easily than an increase of one on thirty .sx This , however , cannot be a complete explanation , for the difference between the standard and the higher mean is greater than it should be by Weber's Law .sx Similar results were given with the modified Heilige colorimeter .sx This instrument consists of a long glass wedge filled with a solution of potassium bichromate .sx This wedge is contained in a wooden box and can be moved upwards or downwards by turning a screw .sx In the side of the box is a small hole fitted with an eye-piece :sx looking through this , one sees a small part of the wedge and , as this is moved up ( or down ) , the depth of fluid and so the intensity of the colour seen through the eye-piece changes .sx The field is so small that its coloration is practically uniform .sx The procedure with this instrument consisted in setting the pointer at the mark 50 on the scale ( by means of which the position of the wedge could be read ) and instructing the subject to look carefully at the colour and remember it , then while his eyes were closed the wedge was moved till the pointer was at scale reading 0 ( giving a white field ) , and the subject then turned the screw until he reached what seemed to him to be the original shade .sx The pointer was again set at 50 , and the subject was told to note the shade , but was not told that it was the same as was given before :sx it was then set at 100 ( the darkest shade ) and he brought it back to apparent equality with the first .sx Twenty-three subjects took part in this experiment ; 19 of them had had a prolonged training in a school of art and were skilled in painting and design .sx Each subject made at least four adjustments , beginning twice at 0 and twice at 100 , in the order 0 , 100 , 0 , 100 .sx The mean of the 41 readings of the artists when they passed from 100 ( the darker ) to the standard was 55.15 ( standard deviation 5.8 ; probable error of the mean ) .sx The mean of the adjustments made in passing from 0 ( the lightest ) was 34.5 ( S.D. 7.2 ; P.E. ) .sx In only 5 of these 82 readings did the adjustment reach or exceed the standard .sx These artists , then , in spite of ( or is it because of ?sx ) their professional interest in colour , show the persistence effect quite clearly :sx indeed , there is a suggestion that they show it more strongly than the layman , for the means of the readings given by the four who had had little or no artistic training were 55.2 ( S.D. 2-62 ; P.E. 0.88 ) in the 100 to standard series and 40-5 ( .sx D. 2.87 ; P.E. 0.97 ) in the other .sx This is analogous to what we found in sound :sx musicians show it very clearly in the tempo of their playing .sx The suggestion made in the previous article applies here :sx the artist being more interested in self-expression than in adjustment to his environment , his behaviour exhibits these fundamental psychic characteristics very clearly .sx 3 .sx PERSISTENCE IN REMEMBERING THE PITCH OF A NOTE .sx The investigation of the persistence phenomenon in the sphere of pitch proved particularly difficult and for a long time appeared to be beyond our technique .sx We used a Stern's tone-variator , an instrument which consists essentially of a metal bottle across the top of which a current of air is directed :sx the capacity of the bottle and therefore the pitch of the note can be varied by turning a knob which raises or lowers its base .sx The pitch of the note can be read from a dial on which a hand moves as the base of the bottle is moved .sx The current of air came from a large organ bellows .sx With this apparatus we made a number of attempts to find whether persistence appears in the estimation of the pitch of a note .sx Several methods were tried .sx In one set of observations a constant standard , e.g. a note of 300 vibrations per second , was followed by a constant higher note , e.g. one of 320 V.P.S. , and this by a third either like or slightly different from the first , e.g. a note of 303 .sx P.S. , and the subject had to say whether the third was the same as the first or higher or lower than it .sx In another set the second note was lower than either of the .sx others , e.g. 280 V.P.S. In yet another the first note varied but the second was always higher ( or lower ) than it by a constant amount , e.g. the three might be 300 , 320 , 304 ; or 280 , 300 , 285 .sx In some series the experimenter sounded the standard note , then the variable , and then gradually turned the screw so as to produce a note which gradually approached the standard , and the subject had to say when the original note was sounding .sx In others the subject himself adjusted the screw , but without looking at the dial .sx With these gradually changing notes the judgments were extremely difficult but less so when the subject himself made the adjustment .sx The experiments indicated the presence of persistence , but they were unsatisfactory on account of the great variability of the judgments .sx The impression given by them was that a very large number of experiments would have to be made and that the subject's task would be much easier if the three notes could be given in more rapid succession .sx Memory for pitch appears to be generally very feeble .sx We found too that the apparatus itself was not above suspicion , for the pitch of a note could be altered slightly by varying the pressure on the bellows .sx The difficulties disappeared when we substituted for the tone-variator a dulcitone which had been specially made for one of us for experimental work on sound .sx A dulcitone is essentially a piano in which tuning-forks take the place of strings .sx For experimental purposes the forks have several advantages ; their pitches do not vary , they give a very pure note , and they can be struck in rapid succession .sx The instrument is fitted with dampers after the manner of a stringed piano .sx Its advantages for the purpose of psychological experiment will be obvious to all who have used tuning forks with movable carriers :sx these carriers are adjusted slowly and with difficulty , and one cannot always be certain of the accuracy of the adjustment and so of the pitch of the note .sx In this modified instrument 12 notes were inserted at one end of the key-board ; they had the following pitches :sx 261 , 261.5 , 262 , 263 , 264 , 266 , 268 , 271 , 275 , 280.5 , 287 , 294 vibrations per second :sx the second , third and following notes differed from the first by 0.5 , 1 , 2 , 3 , 5 , 7 , 10 , 14 , 19.5 , 26 , and 33 vibrations per second .sx The lowest ( 261 ) was the middle C on the scale of the rest of the instrument ( which was unmodified ) , the highest ( 294 ) was middle D. There was one octave below this C , and three above it ; so that the conditions were ideal for the purposes of this experiment .sx For these experiments our subjects were students at the Jordanhill .sx Training College , most of them post-graduate students , and many of them musical .sx For each set of observations they were taken in groups .sx Their threshold for pitch was first measured in the following manner .sx They were told that they would hear two notes which might be the same or different and were asked to say whether the second was higher or lower than the first or equal to it .sx Then another pair would be given and so on until a sufficiently large number of observations had been made to indicate the degree of their sensitivity to differences of pitch .sx Each recorded his or her answer on paper .sx The lowest note was sounded with each of the others but in irregular order and this was repeated five times , the arrangement being different in each set .sx In considering the results we neglected the first set , regarding it as a practice series .sx A week later ( in some cases the same day ) came the persistence series .sx These were of four kinds , which we shall call B , C , D , and E. In B and C the middle ( distracting ) note was higher than the first and third , in D and E it was lower .sx In B and D the third note was higher than ( or equal to ) the first , in C and E it was lower ( or equal) .sx These arrangements can be represented graphically as in Fig. 1 .sx The interval between the sounding of two successive notes in each group was about a second .sx The pairs were given in irregular order and the subjects had no indication of the aim of the experiment :sx they were merely asked to say whether the third note in each group was higher or lower than the first or equal to it .sx The results of these experiments are set out in the following tables .sx In these we have included only the records of all those who in the first pitch experiment ( where the notes were given in pairs ) made no mistake in comparing note the pitch of which differed by 10 vibrations per second or more ( less than half a semitone) .sx