SEDIMENTATION I N T I L T E D V E S S E L S (1)

Koreo Kinosita 1
From the Department of Applied Physics, Faculty of Engineering,
Nagoya University, Nagoya~ Japan
Received June 21, 19/~9

I. INTRODUCTION"
Boycott (2) found, in 1920, that the sedimentation of red blood cor-
puscles is apparently accelerated when the test tube is tilted (Fig. 1).
M a n y workers, chiefly medical, have studied this curious phenomenon
(3), among whom we m a y mention Nakamura and Kuroda (4) as
having produced the standard.

Fro. 1. Boycott's effect.

Nakamura's idea is as follows: First, the upper boundary of tile sus-

pended phase, which, at time t = 0, coincided with tile surface of the sus-
pending medium, be at A B at time t (Fig. 2). Now, if the sedimcntation
velocity of the particles be s (which is considered to be identical with the
velocity of descent of the boundary surface when the tube is held vertical),
we should primarily expect that the space marked with arrows of length
sdt be clarified after the lapse of a short time interval dr. In reality, how-
ever, the boundary CD being unstable, t.he particles in the hatched volume
drift to fill up the blank space above CD and form a horizontal boundary
FG. The final result to be observed is, thus, nothing more than a descent
by dh of the boundary from AB to FG. Nakamura assumes in the above
process t h a t the volume A B G F equals the volume of the arrowed space.
1Now at the Department of Physics and Chemistry, Gakushuin University, Tokyo.
525
526 XOaEO KINOSlTA

In the case of a rectangular cylinder shown in Fig. 2, we can write 2

the arrowed volume = {b sec0 + (c - h)tanO}asdt,
and
the volume A B G F = ab sec0 dh,

where a represents the depth of the vessel. E q u a t i n g these and solving for
h, we get
st sin 0
h= (cA-bsecO)(1-e b ) (1)

which, for a given value of t > O, is greater t h a n

h = st. (2)

F o r 0 = 0 °, (l) is reduced to (2).

. . . . . . ~ . . . . . . . . . . . . 0

A h

~e~G'----------"
}+dr
sdt

FIG. 2. Illustration of Nakamura's theory.

The relation described above can also be stated in the form t h a t the
volume clarified in tile lapse of time dt equals the horizontal projection of
the upper boundaries of the suspended phase (including the free surface
and the ones in contact with the u p p e r walls) multiplied b y sdt. The walls
beneath the suspension havc no essential influence, a IIence, the relations
illustrated in Fig. 3.
2Triangular volumes in the corners shown with dotted lines are neglected.
3On a horizontal bottom the particles arc merely piled up; on an inclined wall some
of them adhere and others slide down slowly. In any case, the motion of the suspended
particles is not influenced essentially.
 SEDIMENTATION IN TILTED VESSELS 5~?

Sedimentation velocity, or, strictly speaking, the descent velocity of the

upper boundary of the suspended phase in any vessel can be calculated ac-
cording to this simple principle, and the theory seems to explain the
experimental data satisfactorily; at least, in Nakamura and Kuroda's
experiments on the sedimentation of cattle blood corpuscles, the accord
of lthe theory and the observation is quite splendid.

Fir,. 3. Effect of upper and lower tilted walls.

Nakamura's theory, ho~'ever, is phenomenological, and the mechanism
of the drift of the particles to form a horizontal boundary--which we call
the "leveling action" after Katsurai (5)--remains ambiguous. Nakamura
employed the term "diffusion"; and some workers are trying theoretical
calculations on the basis of this assumption (6, 7). But there is doubt as
to whether the le~,eling is actually due to diffusion, which means the drift
of particles without any essential mass motion of the suspending medium.
The author understands the sedimentation with a well-marked
boundary which we treat here as characteristic of the case where diffusion
is negligibly small. Well-known exponential-type concentration distribu-
tion will be expected otherwise.

I
I
I
I
r

1"4

\
\ I I
\
\
..... t
"'•-•.4¢rn~ .
Fro. 4. Vessel of NH4C] smoke.
An alternative mechanism to be proposed here is a kind of convection,
a hydrodynamic flow in the suspending medium which carries the particles
with itself. Our principal interest in the following experiments was in
verification of this working hypothesis.
52~ KOREO KINOSITA

2. SEDIMENTATION OF AMMONIUM CIILORIDE SMOKE (8)

A m m o n i u m chloride s m o k e was prepared by the usual process, poured
gently into a vessel 4 illustrated in Fig. 4, and its sedimentation observed
in detail.

n(mm)
90

/"

50
/ •

"/"

t(mm}

~0 5 I0

observed

....... calculated (s: 2.0 mm/mtn.)

Fro. 5. Relation between the extent of descent of the boundary

surface h and time t (NII4C1 smoke).

Fig. 5 shows tile relation between the descent of the boundary surface
h and time t, observed for various inclination angles 0. Broken lines were
calculated after Eq. (1), in which s was taken to bc equal to 2.0 m m . / m i n . ,
the observed value of the descent velocity of the boundary surface when
the cylinder was held vertical (0 = 0°). The s m o k e was washed by passing
through water, dried through desiccative and introduced directly into the
4 We at first used a test tube-type vessel of 3 em. diam. with ordinary cork, but found
it inadequate because the inner side of the cork plays the role of a "tilted wall" and
introduces unnece~ary complications.
 SEDIMENTATION IN TILTED VESSI-;LS 5~9

mr~Imln. ~
8 -- //~alc. ~*"\
/ \
l \
I \
/ \
6--
I \
o 11~ \\

~4

I I I I I I I
0 20 40 60 °
0

Fro. 6. Initial velocities of descent of the boundary surface for various inclination
angles 0 (NILCI smoke). Calculated values are derived fi'om Eq. (l).

vessel; hence, the size of the particles was not so uniform, '~and eoagulat.ion
occurred soon after beginning the observation. The upward bend of the
h-t curves at t > 4 ~ 5 rain. m a y presumably be due to this latter effect.
We dare not conclude this because the trend is not clear when 0 = 0 °, but
it wouht not be unreasonable to suppose that the eoagulation is promoted
when 0 > 0 ° by the convection current which, as we shall show later, is
the cause of the leveling aelion in this ease.

t rain. 3 rain. 5 rain.

7 rain. 9 mm II rain.

Fro. 7. Change in tile appearance of the boundary surface (NH4CI ~mokc).

In Fig. 6 the initial values of the descent velocity of the b o u n d a r y

surface estimated from Fig. 5 are compared with (dh/dt)t=o derived from
Eq. (l). Tile value of 0 which makes the latter maximum can be calculated
analytically, and is found to be 41 ° in our case, which is in good agreement
with the obse,'ved result. Thus, N a k a m u r a ' s theory explains the experi-
Mean diameter of tim smoke partMes was roughly estimated at 8u.
530 KOREO KINOSITA

mental results qualitatively very well, but the quantitative values of h

or dh/dt calculated from Eq. (1) arc not as satisfactory.
Fig. 7 illustrates the change in the appearance of the boundary ac-
companying the progress of sedimentation. 6 These sketches will suffice
to show that the leveling is mainly due to a current of air (a kind of con-
vection) in the vessel. We shall briefly discuss the mechanism of this
convection later (§5), but the situation can be roughly understood if one
considers the distribution of the pressure exerted by the suspended system
on the medium (in this case, air): the diminution of the concentration of
particles in the arrowed space in Fig. 2 will makc the magnitude of the
component of the gravitational pressure in the wall-direction decrease near
the upper tilted wall and increase near the lower one, whose difference is
decidedly larger at the bottom than at the upper part of the tube.

3. SEDIMENTATION OF F I N E EMERY POWDER (9)

Fine emery powder for optical shop, of nominal size No. 1200, after
being homogenized by decantation, was suspended in water, poured
gently into vessels like the one shown in Fig. 8, and its sedimentation
observed.

l
i
//=
'/

iI ///
!/

0 5cm

Fro. S. Vessel of emery powder suspension.

The relations of the extent of descent of the boundary surface h vs. t

for various inclination angles are plotted in Fig. 9. Broken curves in the
figure correspond to the calculated values after Nakamura's theory,
putting s = 2.5 mm./min. (the observed descent velocity of the boundary
when O = 0°). The dependence of h-t curves on ~ is quite different from
what was seen in §2 according to the difference in the shape of the vessel.
The general tendency of the curves can be explained by Nakamura's
theory; especially, the accord between the observed and calculated initial
velocities (dh/dt)t.~ is excellent, but the observed h values become greater
than the calculated ones with the lapse of time.
e The h values in Fig. 5 were read at the middle of the boundary surface.
 SEDIMENTATION IN TILTED VESSELS 531

/ZY/, #

/
/
/

40 -- i/ ///
/ //
/
/
/
/,'/,'¢ ,/

20

r (rain.)
P

0 5 I0

observed

calculated {s=2 5 mmlmin.)

FIG. 9. Relation between the amount of descent of the boundary

surface h and time t (emery powder suspension).

The change in the appcarance of the b o u n d a r y a c c o m p a n y i n g tile

advance of sedimentation is as follows: first, a thin wedge-type clarified
layer appears just below the tilted wall, which proves the validity of
N a k a m u r a ' s basic ideas (Fig. 10 (a)); convection then takes place to fill
up this blank space, and the b o u n d a r y descends gradually in a form having
¸

(o) (b]

Fz~. 10. Change in the appearance of the boundary surface (emery powder suspension).
532 KOREO KINOSITA

a protuberance near the tilted wall as shown in Fig. 10 (b). Convection in

this type of vessel is not very clear, but, in ~t test tube-type vessel, the
currcnt is so violent that one can hardly make any measurement.
The appearance of the wedge-type clarified layer, which shall be called
the V-layer hereafter, means t h a t the leveling occurs with a time-lag be-
hind the sedimentation, which may be one of the causes of the discrepancy
between the actual observation and N a k a m u r a ' s theory. The V-layer
was not distinctly recognized in the case of NH4C1 smoke (prcsumably
because the definition of the boundary was inferior), but we can point out
several cases in which we could just perceive it.
Our original purpose of establishing the validity of the hypothcsis that
the leveling action in Boycott's effect is not due to diffusion but to a kind of
convection current in the suspending medium seems to have been ac-
complished by these experiments. But it is desirable to confirm this again
in the case of erythrocyte sedimentation, because the fact that the sedi-
mentation is much slower in the latter case m a y aid the diffusion theory.

4. S]~DIMENTATION OF ]{ED BLOOD CORPUSCLES

Twenty-five hundredths to 30% of rabbit's blood was dissolved either

in physiological saline or in H a y c m ' s solution, and the sedimentation in
small glass vessels like the ones used in the preceding section of this solu-
tion was observed. Approximate values of the descent velocity of the
boundary surface for various inclination angles were 4, 6, 7, and 8.5
mm./hr., respectively, for ~ -- 0 °, 25 °, 40 °, and 50 ° (in the case of 2.5%
solution).

. ' ' : ' ":'~;:"-:="."" 08

~
:..
:. o~v..: .:..: . ~ : . : i . "..

(5 . : ". ".. .. o.4

• o~. • . . . . 02 . - ' " ' . "
F/. .2
. . .'~ .. ~? ~
.~-" " ::
. ; , . ,o9~
q - . . ' . . ".~ ~ . o,
~.... ..

Fro. 11. Levelingconvectionin blood solution (0.5%).

The movement of red blood corpuscles was traced microscopically

(100 X). The microscopc was set horizontally, the vessel being placed in
the position of the object-glass. No special illumination was needed. The
vessel was fixed on a carrier which could be displaced finely in a plane
perpcndicular to the optical axis in both horizontal and vertical directions,
so t h a t we could perform an observation at any desired position.
A thin wedge-type clarified layer (V-layer), similar to the one men-
tioned in the preceding section, appears again in this case. Microscopic
 SEDIMENTATION IN TILTED VESSELS 533

observation reveals an upward movement of the blood corpuscles along

the edge of this layer. The motion is quite orderly, the boundary with the
V-layer never being inva(led by the corpuscles.
A general idea of the moveme.nt of the corpuscles in the whole may be
obtained from Fig. 11, but cat'(; must bc taken as to the interpretation of
this illustration. The currents are unstablc and are liable to be disturbed
by illumination, 7 exccpt the regular currcnts along the boundary of the
V-layer and the bottom (especially the left half) of the vessel. Consc-
quently, the directions and the velocities jotted in this figure are, except
the ones representcd by thick arrows, nothing more than an example
showing the general trend. The currents represented by thick arrows are
quite stable.
mm/sec.

0.05 ~/ X
oV l I
0.5 1.0 mm
Distonce between the
front woll ond the
observotion plane.

FIG. 12. The diminution of current velocitynear the wall due to viscous rcsistance.

The numerals in the figure represent the velocity (mm./sec.) read

with ocular scale. They correspond to the values in a plane near the front
wall. The velocities in the central plane are expected to be larger, as can
be estimated from Fig. 12, which shows the influence of the viscous re-
sistance near tile wall, but actual measurement of them was not possible
because of tile limitations of the instrument.
A number of complementary observations and experimcnts would be
worth mentioning.
I. Currents in tile blood solution caused by sedimentation described
abovc can be visualized by holding a small flake of methyl violet at an
appropriate position in the vessel. A colored stream comes out slowly and
shows invisible current in the suspension. The general trend estimated by
this very simple method agrees with what was illustrated in Fig. I l.
II. As a consequence of tile convection in tile suspension part, hydro-
dynamic flow also takes place in the upper, already-clarified, part of the
N a k a m u r a e m p h a s i z e s t h e unre]iabilit.y of microscopic o b s c r v a l i o n s in reference to
t h e s e points; b u t our results are reproducible so far ~s w h a t we positively m a i n t a i n are
eonccrned. T h e y are also in accordance with t h c o b s e r v a t i o n s by t h e methyl violet method,.
which is to be described later.
534 KOREO I~NOSITA

liquid. It is a very sIow and not always stationary current, which can be
detected by the methyl violet method or by the observation of red blood
corpuscles remaining in the clarified region. (The concentration of the
latter is, of course, very small.)

- . . . I~R* °
- w- -
• .'.'.. .
'.° "...'..'," **.•
.., . . . . ' . . ".

Fro. 13. Convectioncurrent liable to take place in a vertical-wallvessel.

HI. A weak convection is liable to occur, even in a vessel whose walls

are vertical. Unstable as this current is, one can detect it by the methyl
violet method. The general tendency is sketched in Fig. 13. The current is
assuredly duc to sedimentation, because if one fills the same vessel with
water and tries the methyl violet method, nothing more than a straight
downward motion of a colored stream can be seen.

5. SUMMARY AND DlSCL'SSlONS

A. Boycott's effect was observed on the sedimentation of NH4CI

smoke (in air), fine emery powder (in water), and red blood corpuscles
(in diluted blood solution); and it was established that the leveling action
is not due to diffusion but to a mass motion (a kind of convection) of the
suspending medium. The effect of diffusion is very small, if any.
B. The velocity of this con~,ection current is much greater than the
descent velocity of the boundary surface of the suspended phase, which is
usually regarded as the sedimentation velocity (cf. §4). A general idea of
the leveling convection could be obtained from Fig. 11.
C. Nakamura's theory for the velocity of descent of the boundary
surface is qualitatL~ely valid, but not quantitatively satisfactory in all
cases.
According to conclusion A, it is useless to discuss Boycott's effect on the
basis of the equation of diffusion.
As was pointed out in B, the velocity of the leveling convection is
some 10~ times as large as the descent velocity of the boundary surface.
Conscquently, the author is inclined to picture the sedimentation in tilted
~¢essels with a large horizontal vortex turned up in the suspended phase.
The latter, whose upper boundary coincides with the boundary surface
~vhich we observe, flattens itself as times goes on, wherein the concentra-
 SEDIMENTATION IN TILTED VESSELS 535

tion of the particles in it is kept approximately constant b y the excretion

downwards of surplus particles (Fig. 14).
The crux is the mechanism of this vortex. I t is not difficult to compre-
hend it qualitatively, as was mentioned before, by considering the uneven
distribution, caused b y sedimentation, of the pressure which the sus-
pended system exerts on the medium, or, in other words, the nonuni-
formity'of the density of the suspension (cf. §l). But if we go any farther to
discuss the problem with h y d r o d y n a m i c equations, we are beset with
difficulties. If we wish an a t t a c k in the front, our discussions must be
based on Navier-Stokes' equations, as the viscosity of the medium plays
an important role in this phenomenon. Writing down the fundamental
equations is not essentially difficult, if we admit some postulates, but
solving t h e m to arrive at a solution of practical value seems almost
hopeless. The formulation and the solution of the problem is left to the
future.

{a) {b)

Fro. 14. Vortex model of the sedimentation in tilted vessels.

The influence of diffusion was neglected in the above discussions. In

reality, however, it does play a role, but of the second or third order. The
author expects t h a t a kind of convection in a vertical-wall vessel men-
tioned at the end of the preceding section could be reasoned if one could
have carried out theoreetical calculations on the above-stated principle
with further corrections for this diffusion effect.

ACKNOWLEDGMENTS
Observations and experiments in each section of this article were carried out with the
cooperation of one or more of the students of the Faculty of Engineering, Morihei
Hukaya (§2), Eizi Taki (§3), Kozo Isomura (§3), Taizo Tsukada (§§ 3 and 4), and Keizo
Murase (§4), to whoxn the author is verv much indebted. I also wish to express my cordial
thanks to Mr. Akiya Ookawa and Mr. Nobuhiko Saito of the Kobayashi Institute of
Physical Research. The problem was originally suggested by Ookawa, and lively dis-
cussions with them stimulated me very much whenever we had the chance to meet. A
grateful acknowledgment is due to l)r. Taro Suga, the supervisor of our laboratory, and
to Dr. Tominosuke Katsurai of the Scientific Research Institute, Ltd., for their interest
throughout the work and for reading the manuscript.
536 KOREO KINOSITA

REFERENCES
1. A more dct~iled report will appear in Mere. Facully Engin. Nagoya Univ. in the near
future (in English).
2. Bovc(yrr, A. E., Nature 104, 532 (1920).
3. Resum6 of these investigations can be found in Rcf. (4).
4. ~AKA.MURA,H., AND KURODA, K., geijo ,I. Me~t. 8, 265 (1937) (in French).
5. KhTsl,r~b T., Theory of Colloids, 62, Tokyo, 1947 (in J'lpan.).
6. SAITO, N., AXD OKA, S., Kagaku (= Science) 18, 75 (1948) (in Japan.).
7..~]IYATAKE, 0., AII address at a Sectional Meeting of the Physical Society of Japan,
Oct., 1948.
8. Of the sediment-~tion of this aerosol, Tachibana, T., and Terada, H., have observed
Boycott's effect recently [J. Chem. Soc. Japan 68, 2 (1947) (in Japan.~.
9. Simihr observations are reported by JOIINSON, R., "ANDSMYTll, n., Nature 160, 27
(1947) on the sedimentation of potter's bone in a conical flask. However, it sc~cms
that they were ignorant of the relation to Boycott's effect.