432 EXPERIMENT STATION RECORD. I Vol. 41 



flora, just as in those with OS. lamarckiana. This production of twin hybrids 

 is due to the mass mutation of CE. grandiflora into CB. grandiflora ochracea, 

 the typical gametes producing the velutiim, whereas the mutated ones give 

 rise to the Iwta. The fact that there is almost complete analogy between the 

 twins of CB. lamarckiana and those of CE. grandiflora is considered as evidence 

 of a mass mutation into a reZM^mo-producting mutant for CE. lamarcldana. 



Splitting Iwta and splitting rubiennis have been produced only by three large- 

 flowered races, namely, two Californian species. CE. hookerl and CE. franois- 

 cana, and CE. lamarckiana velutitia. In a large number of other cases the 

 hybrids of the GSnotheras, whether twins or monotypic, are constant in their 

 progeny, apart from the rare cases of Mendelian spUtting. The Iwta and rubi- 

 ennis always split into two types, one of which repeats their marks whereas 

 the second resembles the other grandparent. No constant specimens of keta 

 or rubiennis have been found among them. 



Merrillia, a new rutaceous genus of the tribe Citreae from the Malay 

 Peninsula, W. T. Swingle {Philipinne Jour. Sci., Sect. C, 13 (1918), No. 6, pp. 

 335-34S, jils. 2). — Six genera (9 species) of citrus plants containing what are 

 supposed to be the surviving representatives of a once larger group, and now 

 ranging widely in eastern tropical regions, are treated as comprising a natural 

 group inside the tribe Citreae, namely (subtribe) Balsamocitrinse. The Katinga 

 of Slam and the Malay Peninsula is assigned to this group, being the sole 

 representative of the genus newly erected to receive this form (perhaps the 

 most remarkably aberrant of the citrus fruits), which is described under the 

 name MerrilUa caloxylon. 



FIELD CROPS. 



Field technique in determining yields of experimental plats by the square- 

 yard method, A. C. Abny and F. H. Steinmetz (Jour. Amer. Soc. Agron., 11 

 (1919), No. 3, pp. 81-106, flgs. 3). — This paper, a contribution from the Minne- 

 sota Experiment Station, deals with more extensive observations along lines 

 described by Arny and Garber in a previous article (E. S. R., 40, p. 623), to- 

 gether with a discussion of a method equally well adapted to sampling broad- 

 casted and drilled forages and grains on experimental plats. The work was 

 conducted on variously treated tenth-acre fertilizer plats of barley, wheat, oats, 

 and rye on University Farm and on outlying experiment fields located at 

 Waseca, Morris, Grand Rapids, and Duluth. Ten square yard areas of the 

 standing grain were harvested from each plat shortly before harvesting the 

 entire plat. The areas were located not less than 7 ft. within the plat from 

 sides and ends. The yield of grain in grams was ascertained for each square 

 yard, and various combinations of square-yard areas compared with the yield 

 secured by harvesting the entii'o plat. 



A comparison of the probable error for single determinations of the yields 

 from entire plats and from square-yard areas removed from them led to the 

 conclusion that " from relatively uniform standing grain 4 to 5 systematically 

 distributed square-yard areas removed from tenth-acre plats gave approxi- 

 mately the same probable error for yield as harvesting the products of entire 

 plats ; and the probable error for the yield from 10 square-yard areas removed 

 from tenth-acre plats was approximately two-thirds that for the tenth-acre 

 plats from which the square yards were removed. Where the stands of grain 

 were relatively nonuniform, 5 to 10 systematically distributed square-yard 

 areas were necessary to reduce the probable errors to approximately equal those 

 for the yields from tlie tenth-acre plats from which the square-yard areas were 

 removed. These results are strictly applicable for the seasons of 1918 and 1917 



