OCTOBER 13, 1899. ] 
able error decreased from 103° to 357. 
Moreover, the upward rest was longer in 
the first part of the practice than the other 
three phases combined; but at the close of 
the series it was the same as the downward 
rest, thus showing that the greatest gains 
in voluntary activity are those resulting 
from the practice of the weakest and less 
exercised muscles. 
IV. Estimation of time.—A fter a number of 
preliminary tests the intervals, 82 ¥, 100 ¥ 
and 164 ¥ were chosen. The practice lasted 
from 8 to 16 days, on seven subjects. 
The results justify the following con- 
clusions: (1) The estimate of a given in- 
terval varies for different individuals both 
with and without practice. (2) Practice 
on the same interval may cause the varia- 
tion from the given interval to increase 
with one person and decrease with another. 
(3) Time-estimate is a personal factor de- 
pending upon (a) the nature of the person, 
whether of an impulsive or quiet tempera- 
ment, and (0) upon the point of the fixa- 
tion of the attention, whether to the sen- 
sory or the motor side. (4) There is no 
indifference ‘ point’ from which the subject 
does not vary with long continued practice. 
The changes that practice produces in the 
estimation of time are probably due to fix- 
ing the attention on the movement to be 
performed, in which case the estimate is 
‘shortened in accordance with the growth of 
automatic control, or to the sensory side, in 
which case the time-estimate is made longer 
by practice. 
V. Regulated rhythmical action—In ar- 
ranging apparatus for this experiment the 
probable error was found for the Epison 
phonograph to range from 0.2% to 0.7%; 
for the Lupwie kymograph by BaLrzar, 
from 0.2% to 2.0%; for a drum run by an 
Epison motor driven by carefully tended 
Epison-LALANDE batteries, from 0.1% to 
9.3%.° The Pret, marker was found at a 
break of the circuit to have a latent time 
SCIENCE. 
529 
ranging from 1.1 + 0.097 with the magnet 
cores distant from the armature to 14.77 
--0.03° with the cores close to the armature. 
At a make the latent time ranged from 1.87 
+0.17 to 1.37 + 0.5%. With this marker 
the make is nearly as good as the break 
except for its slightly greater irregularity. 
The Drprez marker from VEeRDIN showed 
a latent time at the break of 3.87 + 0.077 
and of 2.57 = 0.647 at the make. Changes 
in the adjusting spring did not make any 
great change in the figures. The probable 
error of the spark records was. found to be 
== 0.25" independent of the speed of the 
drum. 
In beating time in unison with a sounder- 
click each subject had his own constant 
error; this was generally negative; that 
is, the subjects generally beat time be- 
fore the click occurred. With practice the 
negative constant error tended steadily to 
decrease, to become positive and to increase 
positively. The irregularity steadily de- 
creased. 
VI. Free rythmical action.—The seven sub- 
jects were required to beat time without 
any objective signal. The interval chosen 
at the start was unintentionally shortened 
with the progress of the experiment; it was 
also shortened from day to day. The ir- 
regularity decreased in like manner. 
W. SmyTHE JOHNSON. 
PSYCHOLOGICAL LABORATORY, 
YALE UNIVERSITY, 
NEw HAVEN, CONN. 
NATHAN RUSSELL HARRINGTON. 
NarHan R. Harrineton, of Cleveland, 
Ohio, died in the Sudan on July 26, 1899. 
He was acting for the second time as leader 
of the Senff Zoological Expedition, sent out 
under the direction of Columbia University, 
through the liberality of Mr. Chawes H. 
Senff, of New York City. The party con- 
sisted of Mr. Harrington, Instructor in 
Biology in Western Reserve University ; 
