Raymond Pearl 
223 
mouth, but may be at the level of the mouth. In cases of doubt as to where 
the greatest breadth was, a series of transverse diameters were marked off and the 
longest diameter by actual measurement on the card taken as the breadth. It 
was necessary to do this in but few cases, however. 
To prevent any possibility of misunderstanding as to the measurements taken, 
the outline drawings in Fig. 1 are introduced. 
C 
Fia. 1. Diagram to show measurements taken. A, Non-conjugant. B, Pair of conjiigants. ah and 
a'6' = lengtlis. cd and c'd' = breadths. 
Besides the characters length {L) and breadth (B), the length-breadth index 
= 100 BjL, and the difference (in microns) between the lengths of the two indi- 
viduals of the conjugant pair (= ori; ~ -^Boi-yi) the two individuals of a 
random pair of conjugants or non-conjugants were studied. The " difference" was 
taken as always positive, or in other words, without reference to whether A or B 
was the shorter member of the pair. 
The species of Paramecium was the same in all the series, namely P. cau- 
datum. I was in some doubt at the outstart as to whether the Leipzig form 
might not be P. aurelia, but a careful examination of stained specimens left no 
doubt that this form had but one micronucleus, thus agreeing in this, as in 
all the other specific characters, with the form studied at Ann Arbor, namely, 
P. caudatum. 
Finally, as to the calculation of the biometric constants tabulated, it may be 
said that the ordinary methods, now well known, were followed. In all cases, 
unless a specific statement to the contrary is made, it will be understood that 
Sheppard's correction of the second moment was used in getting the standard 
deviation. In determining the constants for the short series B and D, the 
material was not grouped at all. For the units of grouping in the other series 
the reader is referred to the tables in the Appendix to this paper. The constants 
are tabled to three places of decimals in the case of the means, standard devia- 
tions and coefficients of variation, and four places in the case of the coefficients 
of correlation. It will be understood that in the computing decimals were kept 
to a larger number of places. 
There is one point which needs mentioning regarding the method of recording 
the data in the case of the conjugants. In every instance the first individucd 
29—2 
