[Footnote P: For registering the curves a Hering kymograph was used,
with a loop 2-1/2 metres long. The kymograph rested on felt. With
the aid of the Marey model a pneumographic record was taken now of
the thoracic, now of the abdominal, breathing, never both
simultaneously, since this was extrinsic to my purpose, and it would
have made the whole experiment too complex. The time was recorded by
means of the Jacquet chronograph. For purposes of making more exact
measurements the acoustic current interrupter of Bernstein was used,
attuned to 100 vibrations per second. But this necessitated such
rapid revolution of the drum of the kymograph that the curves were
not compact enough for purposes of demonstration. The levers were
all fitted with micrometer adjustments. They wrote tangentially and,
except the one registering the breathing curve, all points lay in
one vertical line. The error of deflection and that due to the
rondure of the writing-surface were both very slight on account of
the comparative length of the levers and the small extent of the
excursions, and for that reason synchronous points lie practically
in one perpendicular. Only the breathing curve has been moved
somewhat to the left, 7.5 millimeters in figures 6 and 7, 2
millimeters in figure 8, 4.5 millimeters in figure 9. (When the
breathing was very profound, as occasionally happened, the error of
deflection would, of course, have to be taken into account.) The
curves here used as illustrations have been reproduced in the exact
size of the originals by the zinco-graphic method, though somewhat
compressed vertically in order to economize space.]

[Footnote Q: My own expressive movements, on the other hand, are as
pronounced as ever. I still find the attempt to suppress them as
difficult now as when I was working with the horse (page 57). I
could not, of course, procure a curve of these movements of my own.]

Their movements, which at first were quite profuse, decreased more and
more, so that in the case of von Manteuffel the percentage of my
successful responses sank from 73% correct responses in 90 tests to 20%
in a total of 20 tests,--and in the case of Schillings from 75-100% to
23% in a series of 35 tests. The curves obtained with von Manteuffel as
subject, which I am here publishing (figures 8 and 15), are, however,
true to his normal habits.