.1. H. 8. Haldank 297 



A U\h]o of vnliioft of , I e~'*(it hHowm that 7 \h <|uit«' ii pmhalilr 



valut* of jr, whilst tht» rhanro^ ii^iiinst an rrrnr a-s lar^'«- as ihaf iii\o|vi<l 

 by x«8 an* alMuit 50 Ui I. 



It wotiKl \ic UKist (i(>sinihK> to cxaiiiiiic all rxtniil )iiikaL(<- data nii 

 th«»!U» lines. aii<l ih't^'nninr whrthrr int-4'^nil valiirs of .r ami y. if |H»H.Hil»l«' 

 of the fonn 2"— 1, lay within t.h«' zonr of |>rol»ahh' error in alxMit half 

 the wuHes. Wen* this found t<» he the ease it wouM he a stron^^ ar^ni 

 inent in favour of the reduplieation theory, if not it WiiuM tend to 

 disprove that the«>ry, at least in its present, form. 



My thanks aix' due to Professor Kdgeworth, F.IIA , for valiiahl*- 

 adviee and criticism. 



Summary. 



1. F'ormulae are given for the proba])le errors of linkage and re- 

 duplication values calculated from the offspring of crosses of tvjMs 

 A Ii . ab X abab, Ab .oB y. ahab, A B . ah x A B . ah, and A h .aBx A l> . n H 



2. A meth(Kl is given by which the best possi])le linkage values 

 may be calculated from the of^s])ring of the latter two crosses, i.e. from /',. 



3. If this meth<Ml is em])lo3'ed, F^ is almost as accurate a means «»f 

 measuring linkage as arc the (►ffspring from F^ x douhle recessive; it is 

 also slightly more sensitive ;is a means for the detection of linkage. 



REFERENX^ES. 



1. ToDHLXTER. Hutonf of the Theory of ProhdhiUt 11^ p. r).')^. 



2. MoiuiAN ami Bridcjfis. " Sex-linked inheritance in Drosopliil.i.' Curnrifo' /„sti 



tution of Wcuthington^ 1910, 



3. Altexburg. Oenetic*, Vol. i. p. 3r)4, 1910. 



4. PuxNKTT. Journal of Genetics, Vol. vi. No. 3, 1917. 



5. liRllXiE-s. American Xaturalixt, Vol. XLvni. p. 521, 1914. 



