760 



SCIENCE 



[N. S. Vol. XXXVI. No. 935 



interesting. Orthogenesis is a progressive 

 variation or biased heredity in a given direc- 

 tion in successive generations. It is not a 

 constant process, but one which is active in a 

 certain characteristic in a particular species 

 for a limited time, while most other charac- 

 teristics are untouched by it. Let us call one 

 of these lines of orthogenetic activity an 

 " orthogenetic run " after the analogy of the 

 " run " on a special line of merchandise which 

 merchants occasionally experience. In such 

 an orthogenetic run the individual steps are 

 often not affected by selection up to a certain 

 point, or only to so slight a degree as to be 

 relatively negligible. But a "run" may 

 eventually reach the threshold of a decisive 

 selective value. Natural selection will then 

 stop the course of the run. The " run " may 

 be eradicated from the species, or when it is 

 too persistent to be eradicated, then the species 

 will be exterminated, as I imagine has taken 

 place in the Irish elk and in the excessively 

 spiny species with which many extinct mol- 

 lusc genera have culminated. 



But the result of such continued checking 

 of inimical runs and the tolerance of other 

 neutral or beneficial runs is to determine to 

 some degree the kinds of " runs " that will 

 arise in the future. Thus " runs " dealing 

 with distribution of hair and hair modifica- 

 tions in great variety have been tolerated or 

 encouraged, so that runs of this kind became 

 commoner. On the other hand, runs affect- 

 ing the chemical constitution of the red blood 

 corpuscle once it had reached the vertebrate 

 standard have been so promptly checked that 

 variation in this characteristic has been 

 checked and a high degree of conservatism re- 

 sulted. 



The analysis of the experiments given has 

 a further value in its bearing on the question 

 of the possible selective value of minute varia- 

 tions. Indisputably there has been too gen- 

 eral a faith in widespread selective values in 

 the past. At present, however, we can see 

 evidences of an unjustifiably extreme reac- 

 tion. In the consideration of selective values, 

 critics have been wont to compare two adja- 



cent classes of the frequency polygon and to 

 descant on the improbability of a selective 

 value. A safer viewpoint is to consider the 

 chances of two classes not immediately ad- 

 jacent. If a selective value is found between 

 some two classes, we may be fairly sure that 

 the selective value will not end abruptly, but 

 become less step by step in passing from one 

 class to the other. It is absurd to expect a 

 high survival rate in one class and a low one 

 in the next. In almost all cases we must ex- 

 pect the survival rate to gradually increase or 

 diminish from class to class. This is the ver- 

 dict of the experiments. It is idle to talk 

 about the impossibility of selective value of 

 minute differences, when it is possible to meas- 

 ure and analyze them. 



The object of this paper has been to develop 

 the method of analysis of natural selection by 

 the construction of survival curves, in order, 

 first, to extract more meaning from the ex- 

 periments already performed, but second, and 

 especially to encourage further selection ex- 

 periments by making it possible to obtain 

 more significance from them. Its publication 

 has been delayed four years that the author 

 might add some applications of the method. 

 Occupation upon another line of research 

 makes its publication necessary now without 

 such results in the hope that they will be sup- 

 plied by others. 



EoswELL H. Johnson 



University of Pittsbukgh 



SPECIAL ABTICLMS 



BOTANICAL EVIDENCE OF THE AGE OF CERTAIN 



OX-BOW LAKES 



In the southeasternmost county of Ar- 

 kansas, near the Mississippi River, there is a 

 crescent-shaped ox-bow lake about 15 miles 

 long, of a type frequent in the flood-plains of 

 large sluggish rivers, known as Lake Chicot. 

 Lake Village, the county-seat of Chicot 

 County, is located on the side farthest from 

 the river, and the railroad from that point to 

 Luna Landing on the Mississippi skirts its 

 northern bank for a few miles. At present the 

 shores of this lake are mostly pastured, but at 



