The species being tried out in Colombia is Geophagus steindachneri. There are two, possibly three, species of Geophagus in Colombia, one in the Atrato-San Juan, the other in the Magdalena basin up to Girardot. One species is found in Panama and other species are found east of the Andes. They are very active, nervous fishes, dodging in among and hugging the sides of rocks. From their general habits, I should judge them effective sight hunters of larvæ, but I fear they will prove too nervous for the work in small containers, frequently disturbed. The young may survive in the hampered environment of the containers. The males carry the eggs and very young in their mouths.

Two species of "Catfishes" with long whiskers are being used in Colombia, Pimelodella chagrensis and Rhamdia seba. These are fishes of the lowlands and quiet pools, living in general in a very different environment from Pygidium. They have flexible barbels, more than half as long as the entire fish, with which they can (theoretically at least), explore any small container with the minimum of muscular effort. Another species of the genus Pimelodella (yuncensis) was tried by Dr. Hanson in Peru but was abandoned.

Dr. Hanson has mentioned two other species, the "Boca chica " and a brackish water species near Cartagena. The "Boca chica " of the Magdalena, Prochilodus magdalenæ, ascends the river in countless millions during winter. They are dried and used by laborers as food. The adult is certainly not suitable for yellow fever work. The young may serve. Another species of the same genus was tried in Ecuador, Prochilodus humeralis, but abandoned.

The brackish water species is possibly Mollienisia sphenops. Another species of the same genus, Mollienisia caucana is abundant in the swamps of the lower Magdalena basin. It is a minute fish that for a priori reasons should make an ideal species to do the work.

Species of the Poeciliidæ, to which Mollienisia belongs, and a family which contributes mosquito eradicators elsewhere, are widely distributed in Colombia. Along the coasts of the Guianas they are especially abundant both in species and numbers.

ZOOLOGICAL LABORATORY,

UNIVERSITY OF INDIANA.

Piabucina panamensis Gill. The "Volador" of yellow fever work in
Colombia.

UN

THE BEHAVIOR OF CENOTHERA NEO-LA MARCKIANA IN SELFED LINE THROUGH SEVEN GENERATIONS.1

BY BRADLEY MOORE DAVIS,

University of Michigan.

I reported in an earlier paper (Davis 19166) the segregation of a hybrid in the F3 generation from the cross Enothera franciscana X Enothera biennis with so many points of resemblance to Enothera Lamarckiana that it seemed to me to be scarcely distinguishable from Lamarckiana as this collective species would be described in systematic botany. I have named this hybrid Enothera neoLamarckiana and have carried it along in selfed line through seven generations. This paper will give (1) The behavior of neo-Lamarckiana over this period of time, (2) The results of crosses to biennis designed to test its purity, and (3) The behavior, when selfed, of the more prominent derivatives from neo-Lamarckiana.

We shall note that the behavior of neo-Lamarckiana in throwing through successive generations certain characteristic types is similar to that of Lamarckiana when it produces with each generation its particular set of "mutants." Morgan (1923, pp. 309, 310) sees little point in this parallelism of behavior because certain wild species of Enothera studied by de Vries, Stomps and Bartlett show behavior of the same kind as that of Lamarckiana. Morgan, however, apparently overlooks the evidence which indicates that these wild species like Lamarckiana are heterozygous and may not be taken as representatives of pure species (Davis 1922). They together with Lamarckiana are illustrations of impure species, of which the genus Enothera seems almost wholly to be composed, impure species which chiefly breed true because for the most part only such gametes are developed and produce vigorous zygotes as reproduce thereby the heterozygous constitution of the parent stock.

The parent species of neo-Lamarckiana were Enothera fran1 "Genetical Studies on Enothera," XII. Papers from the Department of Botany, University of Michigan, No. 204.

ciscana Bartlett from California (Bartlett 1914) and Enothera biennis Linnæus from Holland (Bartlett 1913), and these species have been fully described and figured as to their contrasting characters (Davis 1916a, pp. 206, 207). Their points of resemblance to Enothera Lamarckiana in various particulars have also been shown (Davis 1916a, pp. 210-212). The F1 generations of reciprocal crosses were compared in an earlier paper (Davis 1914, pp. 190, 191) and will not be discussed further. A set of inter and back crosses involving the F1 hybrids and the parent species has also been described (Davis 1917, pp. 170-178).

It is a matter of interest that my selfed line of franciscana, which has now been carried through ten generations, when pollinated by biennis still continues to give the same type of F1 generation as was obtained in the first cross made in 1912 out of which came the vigorous plant neo-Lamarckiana (Davis 1914, pp. 189, 190). This cross in recent years has been repeated three times and always large proportions of the rosettes have been pale green and weak, with mortality so high that it is difficult to bring even the strongest plants to maturity.

1

This account will begin with a plant 14.53c in an F2 generation. The culture 14.53 was one of five parallel F2 generations from five sister plants of the F1 culture 13.35, franciscana B × biennis (Davis 1914, pp. 190, 191). The five sister F1 plants (13.35ac, 13.35b, 13.35c, 13.35g and 13.35r) differed from one another as has been noted (Davis 1916a, p. 229) and gave F2 generations with various peculiarities (Davis 1916a, pp. 228-242). The plant 14.53c was selected as most likely to give in succeeding generations some Lamarckiana-like offspring.

THE F2 GENERATION.

The F2 generation, culture 14.53 from the F1 plant 13.35€, was from a sowing in earth of 454 seeds from 5 capsules; 387 seedlings appeared from which 376 rosettes developed, and 363 plants reached maturity. The culture as rosettes separated into two groups, 108 plants with obtuse pointed leaves and 268 plants with acute pointed leaves. From the rosettes with obtuse pointed leaves came broadleaved plants with small flowers of the biennis type in having stigmas below the anther tips; a single hero plant appeared from this group