EQUATION IN BREEDING 1 XI'I KIMI.XTS. 365 



on the theory of errors of observation, particularly as related 

 to a-ironomy, phy-ic-. and like subject-, yet so late as H)O2 

 iV.ir-on 1 d'-mon-trated in a mo-t convincing manner that much 

 of the then <-urrently accepted theory wa- \\ron-. and that all 

 of it quite ..\erlooked a factor which miijit be exceedingly 

 important, namely, correlation of judgments. 



The pre-em -tudy i- by no mean- a complete in\ e-i i^at ion of 

 tin- proltlem of per-on.il equation in Mendelian work. (Correla- 

 tion of error- mi-lit t., U- -tndi-d. and certain other matter- ELS 

 well. But the pre-eiit material i- -tati-ticallv entirely inade- 

 quate fr the di-cu--ion of these point-, and it dor- not -rein 

 olli-n more extensive data, -in* in\ < >\\ e- too 



it a tn on the time and good nature of bu-y worker-. 



I null. -r tlr material here presented brings out clearly the pri- 

 maril\ -iiiial ]>oints. It shows that in .1 M'lidelian ratio tin- 



il eijuaiion of the observer mark- a -our. rror \\hich 



in iln- il mai/e i- of considerable magnitude. Tin- >ource 



r quit- hadows in magnitude, in thi- error 



due to random -amplin.. \'et it is the latter alone \\liich i- 



ordinariK considered I .lelianwc '!'!) probaM. -nor 



i Mendi liali latio a- i oiuiinmly c.ilculated tell- one the pmb- 



aliilii\ that the sample < omite<l i^ l( tme representation "I tin- 

 population from \\ hich it \\ a- dra\\ n. It tell- one no; Inn- 

 \\ hale\. i -about the unconscious bia- of t lie . -oiuiter a- a l'a< tor 

 in prodiiein- : he result set down. 



l'.\ \\a\ o| -uminary it may be >aid that in thi- paper i\ idem e 

 i- pn-eiited \\hich shows that: 



i. The ob-er\ed \ : 2 Mendelian ratio- determined from the 

 same foul n - "I mai/e b\- fifteen competent ol , all dilti r 



I'n mi on. aiioth- 



j. The failure of all observers to agree in their di-tribution of 

 kiin.l- into -t \eral categories n--uh- from t\\o causes, \i/.. 

 the e\i-tenre of -..mat i( all\- inn-rmei liate kernel-, and (6) the 



onal bias or idiosyncracj "i iheob-er\er. 



3. The magnitude of the difference- betueen the -e\eral 

 ob>ei-\er- i- -ucli a- to denum-irate that tin- per-oiial equation 



'IV.u-'in ; ;!n- M.iili' mi-lit, with 



S|M . ial !' Phil. I A, Vol 



pp. 



