On Small Differences in Sensation

By Charles Sanders Peirce & Joseph Jastrow (1885)
Classics in the History of Psychology An internet resource developed by Christopher D. Green , ISSN 149 !"#1" ($et%rn to inde&)

On Small Differences in Sensation By Charles Sanders Peirce & Joseph Jastrow (1885) 'irst p%(lished in Memoirs of the of Sciences, 3, #"!8") Presented Posted Jan !!" *ditor+s note, -han.s to Joseph /) $ansdell o0 0or pro1idin2 3e with an electronic 1ersion o0 this te&t) !cd2!

-he physiolo2ical psycholo2ists ass%3e that two ner1e e&citations ali.e in 4%ality will only prod%ce distin2%isha(le sensations pro1ided they di00er in intensity (y an a3o%nt 2reater than a 0i&ed ratio) -he least percepti(le di00erence o0 the e&citations di1ided (y hal0 their s%3 is what they call the #nterschiedssch$elle) 'echner516 2i1es an e&peri3ent to pro1e the 0act ass%3ed7 na3ely, 8e 0inds that two 1ery di3 li2hts placed nearly in line with the ed2e o0 an opa4%e (ody show (%t one shadow o0 the ed2e) It will (e 0o%nd7 howe1er7 that this pheno3enon is not a clearly 3ar.ed one7 %nless the li2hts are nearly in ran2e) I0 the e&peri3ent is per0or3ed with lateral shi0tin2 o0 one o0 the li2hts7 and with a .nowled2e o0 the e00ects o0 a telescope %pon the appearance o0 terrestrial o(9ects at ni2ht7 it will (e 0o%nd 1ery 0ar 0ro3 concl%si1e) -he conception o0 the psycholo2ists is certainly a di00ic%lt one to sei:e) ;ccordin2 to their own doctrine7 in which the o(ser1ed 0acts see3 0%lly to (ear the3 o%t7 the intensity o0 the sensation increases contin%o%sly with the e&citation7 so that the least increase o0 the latter 3%st prod%ce a correspondin2 increase o0 the 0or3er) ;nd7 indeed7 the hypothesis that a contin%o%s increase o0 the e&citation wo%ld (e acco3panied (y s%ccessi1e discrete incre3ents o0 the sensation7 2rat%ito%s as it wo%ld (e7 wo%ld not (e s%00icient to acco%nt 0or a constant #nterschiedssch$elle) <e are there0ore 0orced to concl%de that i0 there (e s%ch a pheno3enon7 it has its ori2in7 not in the 0ac%lty o0 sensation7 (%t in that o0 co3parin2

nown that the two are two stated press%res7 and the 4%estion presented 0or the decision o0 the o(ser1er is7 which is which@ 'ro3 the pro(a(ility7 th%s ascertained7 o0 co33ittin2 an error o0 a 2i1en 3a2nit%de7 the pro(a(le error o0 a 9%d23ent can (e calc%lated accordin2 to the 3athe3atical theory o0 errors) I07 now7 we 0ind that when the ratio o0 the two press%res is s3aller than a certain ratio7 the erroneo%s 9%d23ents n%3(er one!hal0 o0 the whole7 while the 3athe3atical theory re4%ires the3 to (e sensi(ly 0ewer7 then this theory is plainly dispro1ed7 and the 3a&i3%3 ratio at which this pheno3enon is o(ser1ed the so!called #nterschiedssch$elle) I07 on the other hand7 the 1al%es o(tained 0or the pro(a(le error are the sa3e 0or errors 1aryin2 0ro3 three ti3es to one! 0o%rth o0 the pro(a(le error (the s3allest 0or which it is easy to collect s%00icient o(ser1ations)7 then the theory o0 the 3ethod o0 least s4%ares is shown to hold 2ood within those li3its7 the pres%3ption will (e that it e&tends still 0%rther7 and it is possi(le that it holds 0or the s3allest di00erences o0 e&citation) B%t7 0%rther7 i0 this law is shown to hold 2ood 0or di00erence so sli2ht that the o(ser1er is not conscio%s o0 (ein2 a(le to discri3inate (etween the sensations at all7 all reason 0or (elie1in2 in an #nterschiedssch$elle is destroyed) -he 3athe3atical theory has the ad1anta2e o0 yieldin2 conceptions o0 2reater de0initeness than that o0 the physiolo2ists7 and will th%s tend to i3pro1e 3ethods o0 o(ser1ation) /oreo1er7 it a00ords a ready 3ethod 0or deter3inin2 the sensi(ility or 0ineness o0 perception and allows o0 a co3parison o0 one o(ser1er+s res%lts with the res%lts o0 others= 0or7 . (1) >%r 1al%e relates to the pro(a(le error or the 1al%e 0or the point at which an error is co33itted hal0 the ti3e= ( ) in o%r e&peri3ents there were two opport%nities 0or 9%d2in27 0or the initial wei2ht was either 0irst increased and then di3inished7 or 1ice 1ersa7 the s%(9ect ha1in2 to say which o0 these two do%(le chan2es was 3ade) It wo%ld see3 at 0irst (l%sh that the 1al%e th%s o(tained o%2ht to (e 3%ltiplied (y A (1)414) to 2et the error o0 a sin2le 9%d23ent) Bet this wo%ld hardly (e correct7 (eca%se the 9%d23ent7 in point o0 0act7 depended al3ost e&cl%si1ely on the sensation o0 increase o0 press%re7 the decrease (ein2 0elt 1ery 3%ch less) -he ratio A (1)414) wo%ld there0ore (e too 2reat7 and 1) wo%ld perhaps (e a(o%t correct) -he ad1anta2e o0 ha1in2 two chan2es in one e&peri3ent consists in this.nowin2 the n%3(er o0 errors in a certain n%3(er o0 e&peri3ents7 and acceptin2 the concl%sions o0 this paper7 the calc%lated ratio to the total e&citation o0 that 1ariation o0 e&citation7 in 9%d2in2 which we sho%ld err one ti3e o%t o0 0o%r7 3eas%res the sensi(ility) Incidentally o%r e&peri3ents will a00ord additional in0or3ation %pon the 1al%e o0 the nor3al a1era2e sensi(ility 0or the press%re sense7 which they see3 to 3a.ccordin2ly we 0ind that the 0re4%encies o0 errors o0 di00erent 3a2nit%des 0ollow the pro(a(ility c%r1e7 which is the law o0 an e00ect (ro%2ht a(o%t (y the s%3 o0 an in0inite n%3(er o0 in0initesi3al ca%ses) -his theory7 howe1er7 does not ad3it o0 an #nterschiedssch$elle) >n the contrary7 it leads to the 3ethod o0 least s4%ares7 accordin2 to which the 3%ltiplication o0 o(ser1ations will inde0initely red%ce the error o0 their 3ean7 so that i0 o0 two e&citations one were e1er so little the 3ore intense7 in the lon2 r%n it wo%ld (e 9%d2ed to (e the 3ore intense the 3a9ority o0 ti3es) It is tr%e that the astrono3ers the3sel1es ha1e not %s%ally s%pposed that this wo%ld (e the case7 (eca%se (apart 0ro3 constant errors7 which ha1e no rele1ancy to the present 4%estion) they ha1e s%pposed this e&tre3e res%lt to (e contrary to co33on sense) B%t it has see3ed to %s that the 3ost satis0actory co%rse wo%ld (e to s%(9ect the 4%estion to the test o0 direct e&peri3ent) I0 there (e a least percepti(le di00erence7 then when two e&citations di00erin2 (y less than this are presented to %s7 and we are as. I0 only one chan2e were e3ployed7 then so3e o0 the e&peri3ents wo%ld ha1e an increase o0 e&citation only and the others a decrease only= and since the 0or3er wo%ld yield a 0ar 2reater a3o%nt o0 sensation than the latter7 the nat%re o0 the res%lts wo%ld (e 2reatly co3plicated= (%t when each e&peri3ent .ed to 9%d2e which is the 2reater7 we o%2ht to answer wron2 as o0ten as ri2ht in the lon2 r%n) <hereas7 i0 the theory o0 least s4%ares is correct7 we not 5p) #?6 only o%2ht to answer ri2ht o0tener than wron27 (%t we o%2ht to do so in a predicta(le ratio o0 cases)5 6 <e ha1e e&peri3ented with the press%re sense7 o(ser1in2 the proportion o0 errors a3on2 9%d23ents as to which is the 2reater o0 two press%res7 when it is .e a 0iner sense than it has hitherto (een (elie1ed to (e) B%t in this re2ard two thin2s ha1e to (e noted.sensations) In short7 i0 the pheno3enon were esta(lished7 we sho%ld (e 0orced to say that there was a least percepti(le di00erence o0 sensation !! a di00erence which7 tho%2h e&istin2 in sensation7 co%ld not (e (ro%2ht into conscio%sness (y any e00ort o0 attention) B%t the errors o0 o%r 9%d23ents in co3parin2 o%r sensations see3 s%00iciently acco%nted 0or (y the slow and do%(tless co3plicated process (y which the i3pression is con1eyed 0ro3 the periphery to the (rain= 0or this 3%st (e lia(le to 3ore or less accidental deran2e3ent at e1ery step o0 its pro2ress) .

e7 and the lia(ility to error is constant)5"6 -hro%2ho%t o%r o(ser1ations we noted the de2ree o0 con0idence with which the o(ser1er 2a1e his 9%d23ent %pon a scale o0 0o%r de2rees7 as 0ollows.) -he a1era2e 3ar.s see3 to con0or3 to the 0or3%la!! m D c lo2 (pE1!p) where m denotes the de2ree o0 con0idence on the scale7 p denotes the pro(a(ility o0 the answer (ein2 ri2ht7 and c is a constant which 3ay (e called the inde& o0 con0idence) -o show that this 0or3%la appro&i3ates to the tr%th7 we co3pare it with the a1era2e 3ar.s assi2ned to esti3ates o0 di00erences 0or which 3ore than a h%ndred e&peri3ents were 3ade) /r) Jastrow+s e&peri3ents are separated into 2ro%ps7 which will (e e&plained (elow) . C denoted a(sence o0 any pre0erence 0or one answer o1er its opposite7 so that it see3ed nonsensical to answer at all) 1 denoted a distinct leanin2 to one alternati1e) denoted so3e little con0idence o0 (ein2 ri2ht) " denoted as stron2 a con0idence as one wo%ld ha1e a(o%t s%ch sensations) <e do not 3ean to say that when :ero was the recorded con0idence7 there was a(sol%tely no sensation o0 pre0erence 0or the answer 2i1en) <e only 3ean that there was no sensation that the o(ser1er noticed when attendin2 to his 0eelin2s o0 this sort as closely as he con1eniently co%ld7 na3ely7 closely eno%2h to 3ar.e3(races a 5p) ##6do%(le chan2e this di00erence in the a3o%nt o0 sensation ca%sed (y an increase and decrease o0 press%re a00ects e1ery e&peri3ent ali. the3 on this scale) -he scale o0 con0idence 0l%ct%ated considera(ly) -h%s7 when /r) Jastrow passed 0ro3 e&peri3ents %pon di00erences o0 wei2ht o0 ?C7 "C7 and 15 on the tho%sand to di00erences o0 C7 1C7 and 5 on the tho%sand7 altho%2h the acc%racy o0 his 9%d23ents was decidedly i3pro1ed7 his con0idence 0ell o00 1ery 2reatly7 owin2 to his no lon2er ha1in2 the sensation prod%ced (y a di00erence o0 ?C present to his 3e3ory) -he esti3ations o0 con0idence were also ro%2h7 and 3i2ht (e i3pro1ed in 0%t%re wor.

+cd*+6 .5p) #86 -he 9%d23ents en%nciated with any 2i1en de2ree o0 con0idence were 3ore li.s in /r) Jastrow+s second7 third7 and 0o%rth 2ro%ps)546 5%ditor&s note' (he table immediately belo$ is misplaced $ithin footnote ) in the ori*inal publication.ely to (e ri2ht with 2reater di00erences than with s3aller di00erences) -o show this7 we 2i1e the 0re4%ency o0 the di00erent 3ar.

ilo2ra3 placed in the pan o0 the (alance (ro%2ht a press%re o0 one!0o%rth o0 its wei2ht %pon the 0in2er) -he di00erential press%re was prod%ced (y lowerin2 %pon the pan o0 the (alance a s3aller pan into which the proper wei2hts co%ld (e 0ir3ly 0i&ed= this little pan had its (otto3 o0 cor.7 and was placed %pon a piece o0 0lannel which constantly re3ained in the pan o0 the (alance) It was li0ted o00 and on (y 3eans o0 a 0ine India!r%((er thread7 which was so 3%ch stretched (y the wei2ht as certainly to a1oid any noise or 9ar 0ro3 the 3o3ent%3 o0 the descendin2 pan) . . .-he apparat%s %sed was an adaptation o0 a FF post!o00ice scale= %pon the end o0 the (ea3 o0 which was 0i&ed a s4%are enlar2e3ent (a(o%t one!hal0 inch s4%are)7 with a 0lat top7 which ser1ed to con1ey the press%re to the 0in2er in a 3anner to (e presently descri(ed) -his was ti2htly co1ered with an India!r%((er cap7 to pre1ent sensations o0 cold7 etc)7 0ro3 contact with the 3etal) .t 0irst we always (e2an and ended with the hea1iest7 (%t at a later period the plan was to (e2in on alternate days with the li2htest and hea1iest) <hen we (e2an with the hea1iest 5 o(ser1ations 556 were 3ade with that= then 5 with the 3iddle one7 and then 5 with the li2htest= this constit%ted one!hal0 o0 the sittin2) It was co3pleted (y three 3ore sets o0 57 the order o0 the wei2hts (ein2 re1ersed) <hen we (e2an with the li2htest the hea1iest was %sed 0or the third and 0o%rth sets) In this way 15C e&peri3ents on each o0 %s were ta.en at one sittin2 o0 two ho%rs) .e o00 the entire press%re 0ro3 the 0in2er at the end o0 each e&peri3ent) -his wei2ht co%ld (e applied or re3o1ed (y 3eans o0 a ca3 actin2 %pon a le1er= and its (earin2s %pon the (ea3 were 2%arded (y India!r%((er) It was 0o%nd that the %se o0 this arran2e3ent7 which re3o1ed all annoyin2 irre2%larities o0 sensation connected with the re3o1al and replace3ent o0 the 2reater (initial) press%re7 rendered the res%lts 3ore %ni0or3 and di3inished the pro(a(le error) It also shortened the ti3e necessary 0or per0or3in2 the e&peri3ents7 so that a series o0 5 e&peri3ents was concl%ded (e0ore the e00ects o0 0ati2%e were noticea(le) It 3ay (e 3entioned that certain ca%ses tended to the constant decrease o0 the pro(a(le error as the e&peri3ents went on7 these 3ainly (ein2 an increased s. s%00icient wei2ht co%ld also (e h%n2 on the (ea3 o0 the (alance7 so as to ta.ill on the part o0 the operator and an ed%cation o0 the sensi(ility o0 the s%(9ect) -he 0in2er was s%pported in s%ch a way as to (e li2htly (%t 0ir3ly held in position7 all the 3%scles o0 the ar3 (ein2 rela&ed= and the India!r%((er top o0 the (rass enlar2e3ent at the end o0 the (ea3 o0 the (alance was ne1er act%ally separated 0ro3 the 0in2er) -he pro9ectin2 ar3 o0 a 0ilter!stand (the hei2ht o0 which co%ld (e ad9%sted) with so3e attach3ents not necessary to detail7 2ently pre1ented the 0in2er 0ro3 3o1in2 %pwards %nder the press%re e&erted (y the wei2ht in the pan) In the case o0 /r) Peirce as s%(9ect (it 3ay (e noted that /r) Peirce is le0t!handed7 while /r) Jastrow is stron2ly ri2ht!handed) the tip o0 0ore0in2er7 and in the case o0 /r) Jastrow o0 the 3iddle 0in2er7 o0 the le0t hand were %sed) In addition7 a screen ser1ed to pre1ent the s%(9ect 0ro3 ha1in2 any indications whate1er o0 the 3o1e3ents o0 the operator) It is hardly necessary to say that we were 0%lly on 2%ard a2ainst %nconscio%sly recei1ed indications) -he o(ser1ations were cond%cted in the 0ollowin2 3anner.t each sittin2 three di00erential wei2hts were e3ployed) .

sli2ht disad1anta2e in this 3ode o0 proceedin2 arises 0ro3 the lon2 r%ns o0 one partic%lar .ely to select was a1oided) .ilo2ra3 alone %pon the 0in2er o0 the s%(9ect (y 3eans o0 the le1er and ca3 3entioned a(o1e7 and when the s%(9ect said Fchan2eF he 2ently lowered the di00erential wei2ht7 restin2 in the s3all pan7 %pon the pan o0 the (alance) -he s%(9ect7 ha1in2 appreciated the sensation7 a2ain said Fchan2e7F where%pon the operator re3o1ed the di00erential wei2ht) I07 on the other hand7 the color o0 the card directed the wei2ht to (e 0irst di3inished and then increased7 the operator had the di00erential wei2ht already on the pan o0 the (alance (e0ore the press%re was (ro%2ht to (ear on the 0in2er7 and 3ade the re1erse chan2es at the co33and o0 the s%(9ect) -he s%(9ect then stated his 9%d23ent and also his de2ree o0 con0idence7 where%pon the total press%re was at once re3o1ed (y the ca37 and the card that had (een %sed to direct the chan2e was placed 0ace down or 0ace %p accordin2 as the answer was ri2ht or wron27 and with corner indicatin2 the de2ree o0 con0idence in a deter3inate position) By 3eans o0 these tri0lin2 de1ices the i3portant o(9ect o0 rapidity was sec%red7 and any possi(le psycholo2ical 2%essin2 o0 what chan2e the operator was li. -he n%3(ers in the col%3ns show the n%3(er o0 errors in 0i0ty e&peri3ents) <ith the a1era2e n%3(er o0 errors in a set o0 0i0ty we co3pare the theoretical 1al%e o0 this a1era2e as calc%lated (y the 3ethod o0 least s4%ares) -he n%3(er )C51 th%s o(tained in this case (est satis0ies the 3ean n%3(er o0 errors) -he n%3(ers a00i&ed with a si2n denote7 in the %pper row the o(ser1ed (a posteriori) pro(a(le error o0 the 3ean 1al%e as 2i1en7 in the lower row the calc%lated (a .en their places7 the operator was 2o1erned (y the color 5p) 8C6 o0 the s%ccessi1e cards in choosin2 whether he sho%ld 0irst di3inish the wei2ht and then increase it7 or 1ice 1ersa) I0 the wei2ht was to (e 0irst increased and then di3inished the operator (ro%2ht the press%re e&erted (y the . cards sho%ld (e %sed) -hese cards were c%t e&actly s4%are and their corners were distin2%ished (y holes p%nched in the3 so as to indicate the scale o0 n%3(ers (C7 17 7 ") %sed to desi2nate the de2ree o0 con0idence o0 the 9%d23ent) -he (ac. was well sh%00led7 and7 the operator and s%(9ect ha1in2 ta.nowin2 that there wo%ld (e no s%ch lon2 r%ns i0 any 3eans had (een ta. pac.7 or 1ice 1ersa7 so that in the 5C e&peri3ents 3ade at one sittin2 with a 2i1en di00erential wei2ht7 5 red and 5 (lac.s o0 these cards were distin2%ished 0ro3 their 0aces) -hey were7 in 0act7 3ade o0 ordinary playin2!cards) . o0 5 cards were ta.en7 1 red and 1" (lac.t the (e2innin2 o0 a set o0 57 the pac.ind o0 chan2e7 which wo%ld occasionally (e prod%ced (y chance and wo%ld tend to con0%se the 3ind o0 the s%(9ect) B%t it see3s clear that this disad1anta2e was less than that which wo%ld ha1e (een occasioned (y his .)5?6 -he 0ollowin2 ta(les show the res%lts o0 the o(ser1ations 0or each day..t the end o0 each set the res%lts were o0 co%rse entered into a (oo.en to pre1ent the3) .

priori) pro(a(le error) -he last two lines 2i1e the a1era2e con0idence o(ser1ed and calc%lated with each 1ariation o0 the ratios o0 press%re) It will (e seen that the correspondence (etween the real and theoretical n%3(ers is close7 and closest when the n%3(er o0 sets is lar2e) -he pro(a(le errors also closely correspond7 the o(ser1ed (ein27 as is nat%ral7 sli2htly lar2er than the calc%lated pro(a(le errors)5p) 816 -he 0ollowin2 is a si3ilar ta(le 0or /r) Jastrow as s%(9ect.) -he 3ean res%lts 0or the di00erent 2ro%ps are e&hi(ited in the 0ollowin2 ta(les.It wo%ld o(1io%sly (e %n0air to co3pare these n%3(ers with any set o0 theoretical n%3(ers7 since the pro(a(le error is on the decrease thro%2ho%t7 owin2 to e00ects o0 practice7 etc) 'or 1ario%s reasons we can con1eniently 2ro%p these e&peri3ents into 0o%r 2ro%ps) -he 0irst will incl%de the e&peri3ents 0ro3 Gece3(er 1C to Jan%ary 7 incl%si1e= the second 0ro3 Jan%ary 4 to 'e(r%ary 47 incl%si1e= the third 0ro3 /arch 4 to /arch 57 incl%si1e= the 0o%rth 0ro3 /arch "C to the end o0 the wor.5p) 8 6 .

-he ta(les show that the n%3(ers o0 errors 0ollow7 as 0ar as we can con1eniently trace the37 the n%3(ers assi2ned (y the pro(a(ility c%r1e75#6 and there0ore destroy all pres%3ption in 0a1or o0 an #nterschiedssch$elle) -he introd%ction and retention o0 this 0alse notion can only con0%se tho%2ht7 while the conception o0 the 3athe3atician 3%st e&ercise a 0a1ora(le in0l%ence on .

s on pro(a(ilities= Ht is the proportion o0 cases in which the error will (e less than the di00erence (etween the 2i1en e&citations) In all these cases7 o0 co%rse7 we shall answer correctly7 and also (y chance in one!hal0 o0 the re3ainin2 cases) -he proportion o0 erroneo%s answers is there0ore (1!Ht)E ) In the 0ollowin2 ta(le the 0irst col%3n 2i1es the 4%otient o0 the lo2arith3 o0 the ratio o0 e&citation7 di1ided (y the pro(a(le error7 and the second col%3n shows the proportion o0 erroneo%s 9%d23ents.t the sa3e ti3e we 0o%nd the s%(9ect o0ten o1erloo. o0 0i0ty!two once in ele1en ti3es it wo%ld (e necessary to ha1e a sensation a3o%ntin2 to C)"# o0 the pro(a(le error) -his wo%ld (e a sensation o0 which we sho%ld pro(a(ly ne1er (eco3e aware7 as will appear (elow) 5"6 -he n%3(er o0 errors7 when an increase o0 wei2ht was 0ollowed (y a decrease7 was sli2htly less than when the 0irst chan2e was a decrease o0 press%re) .ed this ele3ent o0 his 0ield o0 sensation7 altho%2h his attention was directed with a certain stren2th toward it7 so that he 3ar.-o 2%ess the correct card o%t o0 a pac. Gi1ide the lo2arith3 o0 the ratio o0 e&citations (y the pro(a(le error and 3%ltiply the 4%otient (y C)4##) Call this prod%ct t) *nter it in the ta(le o0 the inte2ral Ht7 2i1en in 3ost wor.psycholo2ical e&peri3entation)586 -he 4%antity which we ha1e called the de2ree o0 con0idence was pro(a(ly the secondary sensation o0 a di00erence (etween the pri3ary sensations co3pared) -he e1idence o0 o%r e&peri3ents 5p) 8"6 see3s clearly to (e that this sensation has no Sch$elle7 and 1anishes only when the di00erence to which it re0ers 1anishes) .7 I7 p) 4 ) 5 6 -he r%le 0or 0indin2 this ratio is as 0ollows.ed his con0idence as :ero) -his happened in cases where the 9%d23ents were so 3%ch a00ected (y the di00erence o0 press%res as to (e correct three ti3es o%t o0 0i1e) -he 2eneral 0act has hi2hly i3portant practical (earin2s7 since it 2i1es new reason 0or (elie1in2 that we 2ather what is passin2 in one another+s 3inds in lar2e 3eas%re 0ro3 sensations so 0aint that we are not 0airly aware o0 ha1in2 the37 and can 2i1e no acco%nt o0 how we reach o%r concl%sions a(o%t s%ch 3atters) -he insi2ht o0 0e3ales as well as certain FtelepathicF pheno3ena 3ay (e e&plained in this way) S%ch 0aint sensations o%2ht to (e 0%lly st%died (y the psycholo2ist and assid%o%sly c%lti1ated (y e1ery 3an) otes 516 *le3ente der Psychophysi.

t 0irst a short pa%se was 3ade in the set o0 57 at the option o0 the s%(9ect= later this was dispensed with) 5?6 In the e&peri3ents o0 Gece3(er7 188"7 and Jan%ary7 18847 the 3ethod as a(o1e descri(ed was not 0%lly per0ected7 the 3ost i3portant 0a%lt (ein2 that the total wei2ht instead o0 (ein2 re3o1ed and replaced (y a 3echanical de1ice7 was ta.2ain7 in connection with a con0idence o0 17 errors occ%rred 1? ti3es in 115 cases (14 per cent))= with a con0idence o0 C7 5 ti3es in 175?C cases ("4 per cent))) 556 . In 171 5 e&peri3ents (s%(9ect7 /r) Peirce) !! 1ariations 157 "C7 and ?C 2ra3s !! there occ%rred con0idence o0 "7 "5 ti3es (" per cent))= o0 7 1C ti3es (9 per cent))= o0 17 8 ti3es ( 5 per cent))= o0 C7 #C? ti3es (?" per cent))) In these e&peri3ents there were "" ( 9 per cent)) errors co33itted7 o0 which 1 (C)" per cent)) was 3ade in connection with a con0idence "= 1C (" per cent)) with a con0idence = 51 (15 per cent)) with a con0idence 1= #C (81 per cent)) with a con0idence C) 'ro3 which we 0ind that in connection with a con0idence o0 " there occ%rred 1 error in "5 cases (" per cent))= with a con0idence o0 7 1C errors in 1C cases (1C per cent))= with a con0idence o0 17 51 errors in 8 cases (18 per cent))= with a con0idence o0 C7 #C errors in #C? cases ("8 per cent))) In 179#5 e&peri3ents (s%(9ect7 /r) Jastrow) !! 1ariations 157 "C7 and ?C 2ra3s !! there occ%rred con0idence o0 "7 ? ti3es (" per cent))= o0 7 19? ti3es (1C per cent))= o0 17 594 ti3es ("C per cent))= o0 C7 171 " ti3es (5# per cent))) In these e&peri3ents there were 451 ( " per cent)) errors co33itted7 o0 which (C)4 per cent)) were 3ade in connection with a con0idence o0 "= 1 (" per cent)) with a con0idence o0 = 9# ( per cent)) with a con0idence o0 1= "4C (#5 per cent)) with a con0idence o0 C) .546 -he res%lt o0 o%r o(ser1ations on the con0idence connected with the 9%d23ents is as 0ollows.en o00 (y the operator pressin2 with his 0in2er %pon the (ea3 o0 the (alance) .2ain7 in connection with a con0idence o0 "7 errors occ%rred twice in ? cases (" per cent))= with a con0idence o0 7 1 ti3es in 19? cases (? per cent))= with a con0idence o0 17 9# ti3es in 5C4 cases (1? per cent))= with a con0idence o0 C7 "4C ti3es in 171 " cases ("C per cent))) In 17?#5 e&peri3ents (s%(9ect7 /r) Jastrow) !! 1ariations 57 1C7 and C 2ra3s !! there occ%rred con0idences o0 "7 none= o0 7 none= o0 17 115 ti3es (# per cent))= o0 C7 175?C ti3es (9" per cent))) In these e&peri3ents there were 5"8 (" per cent)) errors co33itted7 o0 which 1? (" per cent)) occ%rred in connection with a con0idence o0 1= 5 (9# per cent)) with a con0idence o0 C) .

-hese n%3(ers wo%ld (e in closer accordance i0 the pro(a(le error were the sa3e thro%2ho%t7 as it is not owin2 to the e00ects o0 practice7 etc) /oreo1er7 the e&peri3ents were 3ade on di00erent colors !! "CC on white and 1CC each on yellow7 (l%e7 do1e7 pin.ind o0 e&peri3ent in a set and answerin2 accordin2 to that . tho%sand e&peri3ents were 3ade) Gi1idin2 the 3a2nit%de o0 the errors 0ro3 C to the lar2est error7 3ade into 5 parts7 the n%3(er o0 errors7 as o(ser1ed and calc%lated7 that occ%r in each part are as 0ollows.5#6 In the ta(les o0 the third and 0o%rth 2ro%ps7 there is a 3ar.eddi1er2ence (etween the a priori and a posteriori pro(a(le error7 0or thea1era2e n%3(er o0 errors in 5C7 3a.in2 the o(ser1ed pro(a(le error toos3all) -his can only (e partly acco%nted 0or (y the 0act that thes%(9ect 0or3ed the %nconscio%s ha(it o0 retainin2 the n%3(er o0 each.586 -he concl%sions o0 this paper are stren2thened (y the res%lts o0 a series o0 e&peri3ents on the color sense7 3ade with the %se o0 a photo3eter (y /r) Jastrow) -he o(9ect was to deter3ine the n%3(er o0 errors o0 a 2i1en 3a2nit%de7 and co3pare the n%3(ers th%s ascertained with the theoretical n%3(ers 2i1en (y the pro(a(ility c%r1e) .nowled2e)In point o0 0act the pl%s errors and 3in%s errors separately do note&hi(it the sin2%lar %ni0or3ity o0 their s%3s7 0or which we are 4%ite%na(le to acco%nt) -h%s in the 0o%rth 2ro%p we ha1e.7 2reen7 oran2e7 and (rown) -hese e&peri3ents were not contin%o%s) .

Sign up to vote on this title
UsefulNot useful