historical review

Robin Fahraeus: evolution of his concepts

in cardiovascular physiology

HARRY L. GOLDSMITH, GILES R. COKELET, AND PETER GAEHTGENS

McGill University Medical Clinic, Montreal General Hospital Research Institute,
Montreal, Quebec H3G 1A4, Canada; Department of Biophysics, University of Rochester,
Rochester, New York 14642; and Department of Physiology, Freie Universitat Berlin,
D-1000 Berlin, Federal Republic of Germany

GOLDSMITH, HARRY L., GILES R. COKELET, AND PETER one must understand how his ideas differed from those
GAEHTGENS. Robin Fahraeus: evoluti-On of his concepts in car- held by researchers in the early 20th century, the time
diovascular physiology. Am. J. Physiol. 257 (Heart Circ. Physiol. at which Fahraeus began his studies.
26): Hl005-Hl015, 1989.-We give an account of the work of Fahraeus was a pathologist at the University of Upps-
Robin Fahraeus over the years 1917~1938, his contribution to ala in Sweden, and his interest in the suspension stability
our understanding of blood rheology, and its relevance to cir-
culatory physiology. Fahraeus published few original papers on of blood and later in the rheology of blood was motivated
this subject, yet he clearly understood the phenomena occurring by the desire to understand the clinical effects of abnor-
in the tube flow of mammalian blood. 1) The concentration of malities in the aggregation and flow behavior of the
cells in a tube <0.3 mm in diameter differs from that in the formed elements. It is ironic to note, in passing, that the
larger feed tube or reservoir, the Fiihraeus effect. This is due main impact of his work and his ideas were, and are, not
to a difference in the mean velocity of cells and plasma in the in clinical medicine, in particular hematology, but rather
smaller vessel associated with a nonuniform distribution of the in physiology. At the time he wrote his doctoral disser-
cells. 2) In tubes <0.3 mm in diameter, the resistance to blood tation on the suspension stability of the blood (16),
flow decreases with decreasing tube diameter, the Fabraeus- research in the biomedical sciences in Europe was the
Lindqvist effect. We define and generalize the two effects and
describe how red cell aggregation at low shear rates affects cell preserve of researchers interested in clinical problems,
vessel concentration and resistance to flow. The fluid mechan- and it was they who, in the second half of the 19th and
ical principles underlying blood cell lateral migration in tube in the first two decades of the 20th century, had provided
flow and its application to Fahraeus' work are discussed. Ex- a substantial body of results on the pressure-flow relation
perimental data on the Fabraeus and Fahraeus-Lindqvist ef- of blood in glass capillaries (18, 30). The aim of most of
fects are given for red cells, white cells, and platelets. Finally, these studies had been to ascertain whether blood, am-
the extension of the classical Fahraeus effect to microcircula- phibian or mammalian, obeyed the law of Poiseuille (30).
tory beds, the Fahraeus Network effect, is described. One of Poiseuille had provided these investigators with an un-
the explanations for the observed, very low average capillary equaled example of precision and meticulousness in car-
hematocrits is that the low values are due to a combination of rying out the experiments on the flow of liquids through
the repeated phase separation of red cells and plasma at capil- narrow tubes (47). It was Hess (30), in 1915, who put 50
lary bifurcations (network effect) and the single-vessel Fah-
raeus effect. years of work in perspective and concluded (correctly as
it turned out) that blood obeys Poiseuille's law only in
Fahraeus effect; Fahraeus-Lindqvist effect; Network Fahraeus the limit of high flow and shear rates. The non-Newton-
effect; wall effect; axial migration; microcirculation; microves- ian effects, as evidenced by an increased apparent vis-
sel hematocrit; hemorheology cosity at low flow rates were, Hess believed, due to the
elastic deformat ion of the red cells. He thought it plau-
sible that the ability of the red cells to form rouleaux
would play a role in what we would now call the viscoelas-
HISTORICAL
tic behavior of blood. Hess (30) argued against those,
such as Thoma (59), who thought that the decrease in
The hundredth anniversary of the birth (October 15, viscosity with increasing flow rate was due to the exist-
1888) and twentieth anniversary of the death (August ence of a flow-dependent peripheral cell-free layer. Such
18, 1968) of Robin Fahraeus has been recognized, so it a layer had been described as early as the 18th century
seems appropriate to write of his work and to remind by Leeuwenhoek and Boerhaave in capillaries (20) and
ourselves that he holds an important place in the devel- later by Poiseuille in the frog mesenteric microcirculation
opment of modern circulatory physiology. To understand (47). What these pioneers had also seen under the mi-
the significance of his contribution to hemodynamics, croscope was the aggregation of red cells in plasma (20,
0363-6 135/89 $ 1.50 Copyright © 1989 the American P hysiological Societ y Hl005
H1006 ROBIN FAHRAEUS:
47). Strange as it may seem to us today, the aggregation
of red cells and the effect on sedimentation velocity did
not enter the discussion of the non-Poiseuillian flow
behavior of blood in glass tubes. This was so despite the
earlier remarkable observations on the effect of flow rate
on blood cell spatial and velocity distribution in horizon-
tal and vertical tubes by Schklarewsky in 1868 (52).
However, it was generally agreed that the inward migra-
tion of blood cells was due to an inwardly directed
pressure gradient resulting from velocity differences in
the laminae of the sheared suspension, which drives
particles of “high density” (red cells) toward the tube
axis, leaving particles of “low density” (white cells) at
the periphery. Some investigators thought that the
changes in blood viscosity with flow rate were only of
minor significance in the circulation, since “the organism
has a simple means of regulating vessel cross-sectional
area through which the effect of even the most marked
viscosity changes on the total resistance can be compen-
sated for, so that the magnitude of the viscosity coeffi-
cient is quite irrelevant as far as the circulation is con-
cerned” (translated from the German, Ref. 30). Hess (30)
did not subscribe to this view, since he felt that the
relationship between the behavior of blood and the hemo-
dynamics was complex and had not yet been elucidated.
At this point in time, Fahraeus enters the scene, quite
accidentally it appears, in 1917, through his observation
that the sedimentation velocity of red corpuscles in-
creases during pregnancy. It was a well-known fact,
dating back to antiquity, that after blood letting in sick
persons and pregnant women, the buffy coat which
formed during coagulation was absent in the case of
healthy individuals (16, 18). Some had correctly ex-
plained this difference as being due to the greater sedi-
mentation rate of the cells in the “unhealthy blood”; thus
revealing the buffy coat before coagulation stopped fur-
ther sedimentation. However, studies of the problem
ceased in the first half of the 19th century after years of
intense interest. Fahraeus used the question of the buffy
coat as the starting point for his work on red cell sedi-
mentation and the more general problem of the suspen-
sion stability of the blood. He pointed out that fibrinogen
was the principal protein involved in red cell aggregation
leading to the formation of regular rouleaux and that the
process was quite distinct from blood coagulation. He
applied colloid chemical principles to qualitatively de-
scribe the stability of the suspension (16, 18). More
relevant to modern circulatory physiology was the sub-
sequent study of the effect of aggregation on the “stream-
ing blood” (18,19) and the more general relation between
blood cell distribution in vessels and the velocity and
apparent viscosity of blood (21). It is remarkable with
what clarity he understood and how forcibly he expressed
his views on the phenomena occurring in the tube flow
of mammalian blood. The following were his conclusions
(17). 1) At high flow rates in tubes of diameters ~0.3
mm, the concentration of red cells (hematocrit, HT) is
lower than that in the larger feed tube (hematocrit, HF)
because the red cells are distributed in an axial core, and
their mean velocity is therefore greater than the mean
velocity of the blood. There is a reciprocal relationship
between HT and the mean cell velocity in the tube. HT/
1888-1968
HF decreases with decreasing tube diameter. 2) As a
consequence of the lower red cell concentration in the
tube, the measured viscosity in tubes of ~0.3 mm diam-
eter (computed using Poiseuille’s law) is lower than in a
tube of large diameter and decreases with decreasing
tube diameter. 3) The migration of blood cells from the
tube wall toward the tube axis depends on particle size
and not on particle density. Fahraeus thought that Ber-
noulli’s theorem shows that there is a pressure gradient
from wall to axis: the larger the particle, the greater the
force pushing it toward the axis. Hence, it is the large
white cells that are more axially distributed in the tube
than the red cells and whose mean velocity is greater
than that of the red cells. Also, in the advance of a
meniscus of blood into an empty tube, it is the white
cells that accumulate first at the meniscus, forming a
white plug at the blood-air interface. 4) At low flow rates,
the red cells aggregate into rouleaux, and these being the
largest particles in the suspension, migrate to the axis
forming a core that displaces white cells to the periphery.
Therefore, the concentration of white cells in the tube
will be greater than that in the feed tube, and their mean
velocity will be lower than that of the red cells and the
plasma. The formation of an even wider marginal cell-
free zone at low flow rates is expected to decrease the
resistance of flow.
The effect described in the first conclusion is known
as the Fahraeus effect and that in the second conclusion
as the Fahraeus-Lindqvist effect. It has been customary
to associate these with a decrease in cell concentration
and apparent viscosity, respectively. However, it is clear
from the fourth conclusion that Fahraeus realized that
the reverse, an increase in tube particle concentration,
can occur. Some investigators have termed this a reverse
or negative Fahraeus effect. We prefer to include both in
the term “Fahraeus effect,” since, after all, the essence
of the phenomenon lies in a redistribution of suspended
particles, a fact which Fahraeus understood very well.
Moreover, the two effects are also well known in physical
and colloid chemistry (28).
Why then if the effects are so clearly demonstrable
and have such obvious explanations is there so much
literature on both the Fahraeus-Lindqvist effect (begin-
ning with work in the 1940s; Refs. 2, 10, 23) and the
Fahraeus effect (work beginning in the 1960s; Ref. 8)?
There are three reasons for this. 1) The actual experi-
ments carried out by Fahraeus and his students on the
effect of tube diameter on blood cell concentration and
viscosity were surprisingly few in number, and nearly all
data stem from three papers (17, 21, 62). Moreover,
despite the clear assertion that white cells are forced into
the tube periphery by the aggregates of red cells, the
effect of aggregation was never studied quantitatively in
vitro or in vivo. All experiments, whether on cell concen-
tration or viscosity of blood, were carried out at high
pressure gradients, >8 mmHg/cm, at which none of the
low flow rate effects would become apparent. This left
several questions unanswered, in particular whether HT/
HF continues to decrease as the tube diameter approaches
and becomes smaller than that of the red cell. Fahraeus
thought that in a capillary, in which Krogh (37) had
shown that red cells flow in single file, HT/HF = 1 (17).
 ROBIN FAHRAEUS: 1888-1968 H1007

2) The entry into areas of blood rheology and circulatory
physiology of investigators trained in the physical and
engineering sciences, beginning in the late 195os,
spawned a large number of in vitro studies of blood flow
in tubes as well as in Couette and cone-in plate viscom-
eters. 3) The development of sophisticated techniques
for the study of pressure, flow, and hematocrit in the
vessels of the microcirculation in recent years has made
it possible to assess the importance of the Fahraeus and
Fahraeus-Lindqvist effects in the living animal and has
led to conceptual advances, such as the generalization of
the phenomena for microvessel networks as discussed in
DEFINITIONS OF THE EFFECTS and IN VIVO APPLICA-
TIONS.
Before describing the more important advances in the
study of the two effects, we wish to give a brief mathe-
matical definition of the phenomenaand to review the
mechanisms of particle lateral migration in tubes.
DEFINITIONS OF THE EFFECTS
key relationship between parameters. For a given pi,
changing ui forces (HT)i to change in the opposite direc-
tion, e.g., if the velocity increases for a population of
cells (when flux is constant), then their concentration in
the vessel must decrease,. or alternatively, if several
clas ses of cells are flowing through a vessel, each with
the same flux, the fastest cells will be present in the
vessel with the lowest concentration (cell concentration
is inversely proportional to cell velocity).
The total flux of cells through a vessel cross section
can be obtained from Eq. 1 by summing the contributions
of cells of all velocities
C Ni = At C Ui ; (3)
0
The discharge hematocrit is then obtained by converting
the total cell number flux to total volume flux by multi-
plying by Vr and dividing by the total blood flow rate
(AC), where ii is the average blood velocity

F&hraeus Effect v* C N vr
H,, = -=- c UiTZi (4)
Although the change in average concentration of par- iiAAt CAL
titles in a suspension that occurs in flow from a large The vessel hematocrit (HT) can be determined as
into a small diameter vessel is referred to as the Fahraeus
effect, the phenomenon is not to be confused with a C nivr vr
H T- -=-
diminution of particle concentration in the smaller vessel c ni
AL AL
because of an entrance effect whereby particle entry into
the smaller vessel is hindered. To separate this “screen- Dividing Eq . 5 by Eq . 4 gives
ing effect” from the true Fahraeus effect, one needs to -
compare the particle concentration in the suspension in H T
2% 6

the larger feed vessel (hematocrit HF in the case of red -=

cells) with that in the suspension flowing from the
smaller vessel (hematocrit Hn). In steady flow and in the
absence of a screening effect, HF = Hn (24).
Some of the consequences of the Fahraeus effect can
be shown by referring first to the case of steady blood
flow through a cylindrical tube. In such a flow, cell and
plasma velocities decrease with radial distance from the
tube axis, with the highest velocity at the vessel center
if the cell distribution is axisymmetric, but not necessar-
ily so if the cell distribution is not axisymmetric. In
steady flow, the average number of cells (Ni) with veloc-
ity ui crossing a fixed cross-sectional area (A) in the time
interval At is given by the equation
Ni = (UiAt) (1)
where (UiAt) is the axial distance traveled by the cells
with velocity ui in the time interval At, and (ni/L) is the
average number of such cells distributed along the axial
distance L. If both sides of this equation are divided by
A and multiplied by vr, the red cell volume, and if (LA)
is recognized as the vessel volume containing the ni cells,
then we can write
-NiQ _ qi = Ui(HT)i @)
AAtr
where qi is the volumetric flux of such cells across a
vessel cross-sectional area, and (HT)i is the vessel hem-
atocrit of cells with velocity ui. Equation 2 illustrates a
HD
Li -=-
C U;TZi ii,
(6)
since C UiTtLi/~ ni = ii, which is by definition a number
average cell velocity (7). Alternatively, ii, is equal to the
harmonic mean of the velocities of all cells crossing a
fixed cross section of the vessel. In terms of the average
cell and plasma velocities, the equivalent equation to Eq.
6 is
HT
-=- ’ (1 - HT) + HT (7)
HD cc
where i&, is the average plasma velocity. These latter two
equations demonstrate that HT/Hn, which is a numerical
measure of the Fahraeus effect, will change proportion-
ately with changes in ii/i& (from Eq. 6), whereas the
effect of changes in &,/ii, on HT/Hn is more complex,
depending also on the value of HT.
&hraeus (18) set ii/i& equal to HT/Hn, showing that
he had used the equality in Eq. 6 without stating how he
knew that this equality existed.
So far, consideration has been given only to flow in
one vessel. The flow in a vessel can be imagined to consist
of parallel flows in which fluid elements with the same
velocity flow in separate (imaginary) tubes within the
vessel. Such imaginary separation of the flow of fluid
elements does not affect the mathematical description
thus far presented. It is evident, therefore, that the
analysis can be applied to the case of simultaneous flows
in parallel vessels, and furthermore, that the extension
does not require that the vessels be parallel or separate
as long as we consider all the fluid flows that enter or
H1008 ROBIN FAHRAEUS:
leave a volume of tissue completely enveloped by a con-
tinuous surface. These extensions are referred to below.
The limits of the values for HT/Hn depend on the
physics of the situation being considered. In the case of
flow in a single vessel, there are two physical empiricisms
that must be observed: 1) the fluid velocity goes to zero
at the vessel wall, and 2) the apparent viscosity of the
suspension at a given shear rate increases with increasing
cell concentration. As a consequence, in a circular cylin-
drical vessel the minimum value of U/z& is 0.5 for the
case of flow at very low hematocrit in which the cells
flow in single file down the centerline of a large circular
cylindrical vessel; any movement of cells from the cen-
terline results in higher ii/i& values because the velocity
profile in such a flow is parabolic with the maximum
velocity at the centerline. In the case of normal hema-
tocrits, the velocity profile is blunted from the parabolic
because of particle crowding in the suspension (27, 28),
an effect which is further enhanced by the nonNewtonian
flow properties of the blood and by an increased red cell
concentration in the core of the tube. The latter effect is
particularly strong at very low flow rates in vertically
positioned vessels when red cell aggregation leads to the
formation of a centrally located network of rouleaux
surrounded by a cell-depleted peripheral plasma layer.
Here, ii/& lies between 0.5 and 1.0. If the blood flow is
such that the cells are peripherally distributed, in the
low-velocity region, a high value of ii/i& results and is
possibly much greater than unity. This situation can
arise either because of hydrodynamic forces causing a
net migration of cells away from the tube axis [as in the
tubular pinch effect (25, 55) where particles reach an
equilibrium radial position located between tube axis and
wall] or at very low flow rates in horizontal tubes because
of aggregation and sedimentation of cells onto the lower
vessel wall. In the capillaries, if the cell effectively fills
the vessel lumen, ii/i& is very close to unity. Thus, for
single circular cylindrical vessel flows, ii/i& (and there-
fore HT/Hn) is limited to values of 0.5 and larger.
For vessels that do not have a circular cylindrical
lumen, this limit on HT/Hn need not apply. For example,
if a vessel wall is severely convoluted, it is possible that
the fluid regions within vessel wall folds contain only
plasma, which may travel with an average velocity that
is markedly different from the average velocity of the
blood in the central region of the vessel. If all the fluid
volume space is included in the calculation of HT, then
it is possible for HT/Hn to be less than or greater than
05. .
In considering blood flow in networks of vessels, there
are no necessary relationships between velocity and hem-
atocrit in a given vessel; a vessel containing high hema-
tocrit blood can have a high or low flow rate, and the
same is true for a vessel with low hematocrit blood.
Consequently, the ratio of the average hematocrit of all
the blood contained in the network (tissue hematocrit,
HT) to the hematocrit of the blood flowing into the
network (Hn) can have any positive value, including
those ~0.5.
FEhraeus-Lindqvist Effect
The original observation that the flow resistance of
blood and other red cell suspensions decreases as vessel
1888-1968
diameter decreases below -300 pm (21) has come to be
called the Fahraeus-Lindqvist effect.
It has become customary to use the Poiseuille-Hagen
equation to describe the relationship between the vari-
ables that govern blood flow through a cylindrical vessel
Q
- rR4AP
-- (8 a )
h3-L
where Q is the suspension volumetric flow rate through
a tube of radius R under a pressure difference AP over
the vessel length L; qe is an effective viscosity, which
must be empirically determined because the restrictions
used in deriving this equation (laminar flow of a New-
tonian continuum fluid) are usually not met in blood
flow through small vessels. (The term “apparent viscos-
ity” is generally defined as the ratio of shear stress to
shear rate, and for a nonNewtonian or a noncontinuum
fluid flowing in a cylindrical vessel, this ratio varies
across the vessel lumen, making ve some average of the
values of the ratio. For a Newtonian continuum fluid, qe
and the fluid apparent viscosity would be equal.) Re-
arrangement of Eq. 8a gives a means for evaluating ve
from measurable quantities
rR4AP
rle = p W)
8LQ
Often, the effective viscosity is divided by the viscosity
of the suspending medium to give the relative viscosity.
Because the suspending medium (plasma or serum) vis-
cosity is not a function of vessel size or flow rate, the
relative viscosity is proportional to the effective viscosity.
The finding that the relative viscosity of a red cell
suspension decreases as vessel diameter decreases could
be due to a number of factors. It may be simply due to
the Fahraeus effect, i.e., a lower viscosity of the suspen-
sion caused by the lower hematocrit in the vessel; it may
also be due to the rheological effect of a nonuniform
distribution of cells across the vessel lumen or to the
failure of the continuum model of the suspension (see
FLUID MECHANICAL CONSIDERATIONS). The factthatall
three of these factors can be reasonably expected to
contribute to the Fahraeus-Lindqvist effect in blood has
prevented resolution of the question of the relative con-
tribution of these factors to the effect.
Most of the earlier data on the Fahraeus-Lindqvist
effect were obtained under flow conditions effectively
preventing red cell aggregation. However, as pointed out
above, at low shear rates red cell aggregation leads to
different effects. On aggregation, red cells have the tend-
ency to exhibit syneresis so that the mass of red cells
contracts on itself; during flow in a tube, this alone
results in the pulling of the red cells into the center of
the vessel, leaving a larger layer of low-viscosity sus-
pending medium near the vessel wall (resulting in a
lowering of flow resistance). However, aggregated red
cells also have an increased tendency to sediment. Fah-
raeus, of course, was very much interested in the rela-
tionship between red cell aggregation and sedimentation.
Red cell sedimentation in living vessels and its path-
ological significance were extensively studied by Knisely
(36) who used the term “sludged blood” for intravascular
erythrocyte aggregation, which in his view, in clear con-
 ROBIN FAHRAEUS:
tradiction to Fahraeus’ concepts, was an exclusively
pathological phenomenon. The effect of increased red
cell sedimentation on flow resistance depends on the
orientation of the vessel in the gravitational field. In a
vertically oriented vessel, the gravitational effect accel-
erates or decelerates the central core of red cell-rich fluid
depending on flow direction. This has a minor effect on
the flow resistance. Regardless of the flow direction, red
cell syneresis causes an increased axisymmetric suspend-
ing medium layer at the vessel wall that results in a
decreased flow resistance (26, 50). In a horizontal tube,
sedimentation causes a nonsymmetrical distribution of
red cells onto the bottom of the tube, which is found to
result in increased flow resistance (50). (On the basis of
earlier discussion, it is clear that syneresis-sedimentation
effects will also affect the magnitude of the Fahraeus
effect, although this has not been experimentally inves-
tigated yet). The sedimentation causes transient changes
in overall vessel hematocrit, but because of the slowness
of the flow, this effect is seen as a very slow change in
overall flow conditions, although it does depend on the
ratio of the total vessel length to the average fluid veloc-
ity.
FLUID MECHANICAL CONSIDERATIONS
The redistribution of blood cells in narrow tubes,
which is at the root of the Fahraeus and Fahraeus-
Lindqvist effects, is due to a net lateral migration of the
particles away from the vessel wall. Fortunately, there
exists today a substantial body of theoretical and exper-
imental knowledge on the effect of the vessel wall on the
motions of suspended particles in model dispersions and
blood. The reader is referred to reviews of the subject by
Brenner (5) and Lea1 (38). From these studies, carried
out over the last 30 years, it is clear that both the particle
density-dependent hypothesis of Schklarewsky (52) and
the particle size-dependent hypothesis of Fahraeus (17)
for inward migration of blood cells in flow through cir-
cular tubes are false. Both invoke the Bernoulli theorem
to prove that there is a radial pressure gradient across
the tube that results in a force moving particles from
tube wall to axis. The more common expressions of
Bernoulli’s theorem do not account for viscous dissipa-
tion of energy and therefore apply strictly only to an
inviscid fluid and can at best only be used for real, viscous
fluids at high tube Reynolds number when inertial forces
dominate viscous forces. The more complete version of
Bernoulli’s theorem, as well as the application of the
laws of conservation of momentum, show that there is
no radial pressure gradient in the tube. Rather, it is the
wall together with the inertia of the fluid that produces
migration effects, as described below.
Dilute Suspensions
In dilute suspensions of model particles and blood cells
undergoing laminar viscous (Poiseuille) flow through a
circular tube, particle radial migration can occur through
two mechanisms (5, 25, 28, 38). 1) Migration occurs
through lateral movement toward the tube axis of de-
formable particles (liquid drops, flexible fibers, red blood
cells, and rouleaux) at low Reynolds number under creep-
1888-1968 H1009
ing flow conditions (negligible inertial effects). Such
migration persists at higher Reynolds number when in-
ertial effects become significant. In the case of liquid
drops it has been shown to be due to an inwardly directed
force arising from the interaction of the flow field around
the deformed drop with the tube wall (5,6). 2) Migration
also occurs through two-way lateral movement to an
eccentric radial equilibrium position of rigid particles
(spheres, cylinders, and aldehyde-fixed red blood cells)
initially located either near the wall or near the axis, at
higher Reynolds numbers at which inertia of the fluid is
significant. In this, the “tubular pinch effect” (55), mi-
gration becomes appreciable at particle Reynolds number
(Re,) >10e3, where
Re
Here, b and R are the respective particle and tube radii,
0 is the mean linear fluid velocity in the tube, and p and
q are the respective fluid density and viscosity.
The migration velocities in both mechanisms vary as
the third power of the ratio of particle to tube radius,
and the inward migration velocity falls off rapidly with
increasing distance from the tube wall. For migration at
higher Reynolds number, theory shows that the effect
has its origin in inertia of the fluid and in the interaction
of the particle with the wall (5, 12, 38). The effect may
be likened to the curving of a spinning tennis ball (Mag-
nus effect), although it should be noted that nonrotating
particles also migrate, and the presence of the wall is an
essential feature of two-way migration. In Fahraeus’
experiments, conducted at 100-mmHg pressure in a lo-
cm long tube of 0.3 mm diameter, the estimated mean
velocity (0) equals 89 mm/s, assuming Poiseuille’s law
and a high shear rate blood viscosity of 3.5 mPa* s. This
yields a Rep value of 3.8 x 10D5 for red cells (equivalent
sphere radius of 2.8 pm) and 1.1 X 10m4 for the larger
white cells (b = 4 pm). The tubular pinch effect has been
observed with normal as well as glutaraldehyde-fixed red
cells at Rep >10e4 (25), and the migration velocity has
been shown to increase with increasing b/R and 0. White
cells in plasma also migrate appreciably inward from the
tube wall at Rep >10e3 (29). Thus Fahraeus was correct
in believing that the migration effect was particle size
dependent.
The above considerations apply to suspensions of neu-
trally buoyant particles. It is of interest to note that for
nonneutrally buoyant particles, the equilibrium position
for flow in vertical tubes in the tubular pinch effect is
shifted to the wall for a particle sedimenting in the
direction of the flow and is shifted toward the axis for a
particle sedimenting in a direction opposite to that of
the flow (5). The condition for such effects to become
important is that (b/R)2 << u,/o << 1, with u, being the
sedimentation velocity. In the case of Fahraeus’ experi-
ment, u, N 10m2 and 5 x 10m3 mm/s for red and white
cells, respectively; thus u,/U << (b/R)2 in the 0.3- and
even in the O&mm diam tube, with mean fluid velocities
of 50-90 mm/s. The system was thus effectively neutrally
buoyant, and Fahraeus was correct in stating that
Schklarewsky’s density-dependent migration theory was
wrong.
HlOlO ROBIN FiiHRAEUS:
Concentrated Suspensions
It may seen ironic that the law enunciated by Poiseuille
from his work on simple Newtonian fluids (Eq. 8a)
should fail to apply to suspensions of blood cells, since
the object of his research was to arrive at an understand-
ing of flow in living vessels. However, such non-
Newtonian behavior is quite characteristic of many sus-
pensions of large and small particles. Ever since the time
of Bingham it has been known that the effective viscosity
of clays, paints, and other suspensions measured in cap-
illary instruments decreases with increasing flow rate
and decreasing tube diameter (3). To account for these
anomalous effects, three main theories have been pro-
posed.
Hydrodynamic waZZ effect. Vand (61) showed that in
the presence of a rigid boundary a flowing suspension of
rigid spheres of radius b behaves as if there were a layer
of thickness, h, equal to 1.302b at the wall, having a
viscosity equal to that of the suspending medium. The
true (Q) and effective (7;) relative viscosities are related
bY
;-l=($-1)(1-g+ (9)
The theory therefore predicts that ~1 decreases with
decreasing tube radius.
Summation or “Sigma hypothesis. n This is based on
the presumed existence of unsheared laminae within the
suspension that produce a stepwise velocity distribution,
the integral in the derivation of Poiseuille’s equation
being replaced by a summation. This approach is an
empirical one and has been applied to clay pastes and
flowing blood (53), but it has not been possible to relate
it to the problem of dilute suspensions of spheres.
Axial migration of particles. Inward particle migration
across the planes of shear, which as described above
certainly occurs in dilute suspensions, also plays a part
in concentrated suspensions, emulsions, and blood. From
in vivo and in vitro studies at high flow rates (no red cell
aggregation), it appears unlikely that, at hematocrits of
0.40-0.45, the plasma-rich peripheral zone can be much
larger than 4 pm in vessels the diameters of which X00
pm. Nevertheless, an inward migration by only one-half
particle diameter from the wall brings about a significant
decrease in the average particle concentration in the tube
(Fahraeus effect; Refs. 1,56). This results in a significant
decrease in the energy dissipated in the flow and a
decreased effective viscosity, as shown by experiment
and calculation in blood (44, 60) and model particle
suspensions (33, 57).
EXPERIMENTAL DATA
The published experimental data describing the Fah-
raeus and Fahraeus-Lindqvist effects are shown in Figs.
1 and 2, respectively.
Fihraeus Effect
Figure 1A indicates the range of values found for the
Fahraeus effect for the situation in which the hematocrit
in the feed vessel is 0.40-0.45 and the conditions do not
1888-1968
permit red cell aggregation. The only points drawn are
for Fahraeus’ original data (18). It is evident that, at a
given vessel diameter, the data cover a relatively large
range of concentrations. This is in part due to imprecise
experim .entation . and meth .odo logy (e.g. in most cases Hn
was not directly measured but assumed equal to the feed
hematocrit), but it may also be due to a real dependence
of the Fahraeus effect on parameters in addition to vessel
diameter and feed hematocrit, such as shear rate. There
is also the possibility that some of the scatter is due to
differences between experimental conditions, since in-
vestigators used different suspending media (saline and
plasma) and different temperatures (room temperature
and 37-38°C).
The Fahraeus effect does depend on the level of the
feed hematocrit. HT/Hn at a given vessel diameter de-
creases with decreasing HF, the decrease being larger for
smaller vessels.
As mentioned earlier, red cell aggregation under low
flow or pathological conditions will affect the Fahraeus
effect; it is expec ted that movement of cells into higher
velocity (celntral) regions will cause a decrease in HT/Hn,
whereas movement of cells toward lower velocity (wall)
regions will cause an increase in HT/Hn. This has not
been adequately researched.
Figure iB shows the data from human white cells (29,
45,46,62) and platelets (11,58) flowing in blood through
cylindrical tubes. In keeping with the widespread obser-
vations that at high flow rates white cells are preferen-
tially located in the central region of a vessel but that
they migrate to the vessel perimeter as the flow rate
decreases allowing red cell aggregation to occur, the data
in Fig. 1 show that at higher shear rates ( flows) the ratio
of cell number concentration in the tube to that in the
discharge (C&n) is lower than at lower shear rates. The
data from Goldsmith and Spain (29) and Nobis et al.
(46) clearly demonstrate this. With respect to the data
of Nobis et al. (46), it should be poi nted out that these
points were calculated from published experimen tal con-
centration profiles, assumin .g a parabo lit velocity profile.
Because the velocity profile is undoubtedly blunted from
the para boli .c (27), the calcula ted CT/CD va lues underes-
timate the real values.
The platelet data always show CT/CD to be greater
than unity as long as red cells are present; in the absence
of other cells, CT/CD is less than unity for platelets. This
is consistent with the finding that platelet concentration
in blood is greatest near the vessel wall. The datum point
from Tangelder et al. (58) is calculated from concentra-
tion profile data, by the method used for the Nobis et al.
(46) white cell data, and is similarly an underestimate of
CT/CD-
All these data are consistent with the general concept,
applicable to the tube flow of any suspension or emulsion,
that the concentration of particles within the tube de-
pends on their radial distribution. The concentration in
the tube is greater than that in the discharge from the
vessel if the particles are preferentially located in lower
velocity (wall) regions and vice versa.
F&hraeus-Lindqvist Effect
Figure 2 illustrates the Fahraeus-Lindqvist effect for
blood flow through tubes under various conditions. The
 ROBIN FAHRAEUS: 18881968 HlOll

FIG. 1. Fahraeus effect as demonstrated by literature

data. A: Fahraeus effect for human red cells, feed hem-
atocrit 0.40-0.45; all data collected for red cell suspen-
sions flowing through cylindrical tubes at flow rates
ensuring no red cell aggregation. Cross-hatched region
contains all literature data; only points of Fahraeus’
data (18) are shown. Critical diameter (-2.7 pm) is that
of smallest cylindrical vessel through which a human
red cell can flow. HT/Hn, numerical measure of Fah-
raeus effect. B: Fahraeus effect for human white cells
and platelets. CT/CD is ratio of number concentration
1 I, 1 , , , ,,, I 1 ,111ll I I IIll of cells in blood in tube to that in discharge from tube.
2 3 45 10 50 100 loo0 White cell data show expected effect of shear rate (flow
1.5 rate). Data of Vejlens (62) and of Nobis and Gaehtgens
White Celh Platelets
B v Vejlens (1938) Lymphocytes 0 Tangelder (1982)
(45) were obtained at high shear rates. Data of Gold-
A Vejlens (1938) Granulocytes 0 Corattiyl 6 Eckstein (1987) smith and Spain (29) were obtained at mean shear rates
I
Nobis & Gaehtgens (1981) from 48 to 88 s-l for 150-pm tube and from 36 to 118
: Nobis el al. (1982) T ?
n Goldsmith & Spain (1984) s-’ for loo-pm tube, and those of Nobis et al. (46) were
1.3
obtained at mean shear rates from -43 to 330 s-l.
Dashed lines with arrows pointing in direction of in-
creasing shear show how Fahraeus effect changes with
increasing shear rate. Points from Nobis et al. (46) were
1.1 calculated from published concentration profiles with
assumption of a parabolic velocity profile. Platelet data
of Corattiyl and Eckstein (11) show a maximum CT/CD
CT
for an apparent wall shear rate of 800 s-l. In shear rate
CD
range of 80-8,000 s-l, CT/CD decreased from these max-
0.9 imum values to minimum values indicated by lower
circles. Tangelder point (58), obtained at a mean shear
rate -890 s-‘, was calculated from in vivo concentration
profiles, assuming a parabolic velocity profile. All these
CT/CDs must approach unity as vessel diameter de-
0.7 creases.

OS-
2 3 45 10 50 100

Diameter, pm

cross-hatched area contains all the literature data for sistance, as reflected in the relative viscosity, increases
flow rates at which red cell aggregation is prevented with decreasing apparent wall shear rate (flow rate). For
(including all those in Ref. 23 and in all subsequent such a two-phase flow just one curve is shown, which is
data); the only points shown are those from the data of that for wall shear rates of 1.6-4.0 s-l (50).
Fahraeus and Lindqvist (21). As before, differences in At very low flow rates in vertical tubes, the red cells
relative viscosity at a given tube diameter reflect the fact aggregate and migrate into the vessel core, forming a
that experiments were conducted with red cells sus- plasma layer at the wall. The thickness of the peripheral
pended in various media and at different temperatures. plasma layer increases with decreasing flow rate until a
Attempts to predict the observed dependence of rela- constant value is reached. The relative viscosity mirrors
tive viscosity on tube diameter on the basis of the Fah- the growth of the plasma layer, decreasing until the wall
raeus effect, assuming that hematocrit is uniform across layer thickness is constant (26). The lower curves,
the vessel lumen and that fluid viscometric data obtained marked P and D, show the minimum relative viscosities
in Couette and cone-in-plate instruments are applicable, found for red cells in plasma and plasma plus Dextran
have met with varying degrees of success in the hands of 250, respectively (50). In the presence of Dextran 250,
various investigators, both with in vitro and in vivo data. the red cells aggregate more strongly than in plasma
Two-phase flow models with a plasma wall layer and a alone, resulting in a more densely packed central region
core region of uniform or nonuniform hematocrit have of red cells and a thicker wall layer of suspending me-
also been applied to the data and yield values for free dium.
parameters in the models (such as wall layer thickness). For very small vessels, the diameters of which ap-
The calculated parameters often seem physically plausi- proach that of the red cell, no aggregation or sedimen-
ble but do not validate such models. tation is possible, so all curves must converge. In these
At very low flow rates in horizontal tubes, where red smallest vessels, relative viscosity increases dramatically
cell aggregation and sedimentation occur, the flow re- as vessel diameter approaches 2.7 pm, the approximate
H1012 ROBIN FAHRAEUS: 1888-1968
ZO blood viscosity as early as 2 years after the publication
of the Fahraeus-Lindqvist paper, it took about 40 years
until the development of appropriate methods permitted
an evaluation of Fahraeus’ concepts in the living micro-
6.0 circulation. Intravital microscopy, however, had for some
time been used to obtain at least descriptive information
on the effect of red cell aggregation on microvascular
blood flow. Such studies led Knisely (36) to the concept
5.0
of sludged blood as a major cause of increased flow
I u Asymmetric syneresis resistance and eventually even circulatory failure under
pathophysiological conditions. Knisely’s (36) conclusion
that flow stagnation is a consequence of aggregation (or
agglutination as he also called it) was, of course, the
-z?
Ei 4.0 direct opposite of Fahraeus’ conclusions, and a bitter
.-ti controversy developed between the two men. In the light
> of today’s knowledge both Knisely and Fahraeus were
.-z right to some degree. As already pointed out above, red
g 3.0 cell aggregation per se, by causing axisymmetric syner-
fx esis, reduces rather than increases effective viscosity as
Axisymmetric syneresis
was claimed by Fahraeus. However, aggregation in the
presence of gravitation also leads to sedimentation and
nonaxisymmetric syneresis which indeed elevates viscous
2.0
resistance as postulated by Knisely. It is in a way tragic
that the development of our knowledge led to this rather
balanced compromise that leaves room for both alterna-
tive concepts some 20 years after the death of both
1.0 contenders.
I Critical diameter It was Fahraeus himself who stated that the phenom-
enon, now termed the Fahraeus effect, is relevant only
in what he called the “paracapillary vessels” (ID ~300
b 1 I I I pm; Ref. 17). Because the hematocrit in these vessels is
1 10 100 1000 a determinant both of resistance to flow and of oxygen
Diameter, pm transport capacity, quantitative measurements of red cell
FIG. 2. Fahraeus-Lindqvist effect as indicated by literature data for flux and red cell concentration distributions were under-
human blood [hematocrit in feed tubes (Hr) = 0.40-0.451 flowing in taken in various tissues. Such measurements inevitably
cylindrical tubes. Shaded area contains all published data for human demanded an evaluation of Fahraeus’ concepts.
red cell suspensions flowing at sufficiently high rates to prevent red In general terms, the data available today clearly dem-
cell aggregation; points (m) from data of Fahraeus and Lindqvist (21)
are shown. Top curve, from data of Reinke et al. (Ref. 50; l ), indicates onstrate a substantial reduction of hematocrit in the
relative viscosity of blood for very slow flow in horizontal tubes where microcirculation (Fig. 3). This is consistent with the fact
red cell aggregation and sedimentation occur (average wall shear rate that total red cell mass and plasma volume are not
-1.6-4.0 s-l). Relative viscosity increases as average wall shear rate distributed evenly but that the average hematocrit of all
decreases. Bottom 2 curves are for slow flows in vertical tubes, during
which red cells aggregate and draw into central region of tube. Curve the blood in a tissue (tissue hematocrit) is usually less
marked P is for red cells in plasma and denotes minimum relative than large vessel hematocrit. However, considerable
viscosity. Curve marked D is similar, except aggregation was enhanced uncertainty developed concerning the mechanisms con-
by addition of Dextran 250 to plasma. All these curves must converge tributing to the observed hematocrit distribution in mi-
as vessel diameter decreases. crovascular networks. Several aspects emerged that can-
size of the smallest vessel through which a human red not easily be accounted for by the classic Fahraeus effect
cell can travel without rupture. The large rise in the rate in single microvessels alone, and extensions of Fahraeus’
of increase in relative viscosity is primarily due to the concepts have therefore been undertaken. 1) Although,
small clearance between the individual red cell and the on the average, hematocrits appear to decrease with
vessel wall. decreasing vessel diameter, the extent of hematocrit re-
duction is greater than predicted from in vitro measure-
ments. In particular, capillary hematocrit in several tis-
IN VIVO APPLICATIONS
sues was found to average -26-42% of systemic hema-
Fahraeus’ theoretical and experimental work was con- tocrit (31, 32, 34, 35, 41-43, 49, 51, 54; Fig. 3), whereas
ceived as an effort toward a better understanding of the the classical Fahraeus effect alone does not allow ~50%.
blood flow behavior in the circulatory system and partic- The question therefore arose about the origin of this
ularly its “business end,” the microcirculation. Neverthe- “additional hemodilution.” 2) The variability of hema-
less, Fahraeus does not appear to have tested his concepts tocrit values in vessels of similar dimensions within a
by in vivo experiments directly. Although other investi- given microvessel tree appears to increase toward the
gators (63) provided evidence from whole organ perfusion true capillaries, in which most studies have reported
experiments for the diameter dependence of apparent coefficients of variation of -50% (Fig. 3). Because this
 ROBIN FAHRAEUS: 1888-1968 H1013

n I 1 1 I I 1 , Hr/Hn > 0.5 for each vessel. Common to both concepts,

however,is the conclusionof a relative retardation of the
0.8 plasma passing through the vessel network, such that
the ratio of mean transit times ( &/&~,OJ is significantly
smaller than unity.
0.6
Concept A
The value of HT/Hn, as calculated from experimental
data, can be ~0.5 if immobilized plasma volume is in-
cluded in the calculation of Hr as described in DEFINI-
TIONS OF THE EFFECTS. On the one hand, this has been
thought to occur because of noncircular vessel cross
sections, e.g., in contracted arterioles which provide
l Johnson et al. (1971) v Schmid-Sch6nbein & Zweifach (1975) “pockets” of plasma not participating in flow. On the
x Lipowsky & Zweifach (1977) v Sarelius & Duling (1982)
w Lipowsky et al. (1978) o Kanzow et al. (1982) other hand, it has been proposed that a layer of plasma
A Klitzman & Duling (1979)
A Lipowsky et al. (1980)
+
0
Klitzman & Johnson (1982)
Pries et al. (1986)
is immobilized by the roughness of the endothelial sur-
face (“hairy wall”) caused by an extensive glycocalyx and
microvilli. Quantitatively, the thickness of the layer of
B’=°F---n’
0.8 i-
’’’’’’ plasma required to reconcile the low capillary Hr is in
the order of up to 1.2 pm (13, 15,34).
The experimental evidence in support of this concept
includes measurements of Hn in various arteriolar and
0.6 t venular microvessels by direct aspiration of blood
through micropipettes. The data showed values that were
not significantly different from systemic hematocrit, Hsys
(13, 15). In addition, microinjection of heparinase
through micropipettes, presumably causing removal of
glycosaminoglycans from the endothelial surface, re-
sulted in an elevation of capillary hematocrit (14). These
n
observations support the notion that red cell exclusion
0 1 I I I l I I
from an immobilized plasma layer causes Hr/Hn to
80 60 40 20 20 40 60 80
decrease below the limit of 0.5. While red cell flow
Arterial Capillary Venous
fraction (Hn) is constant throughout the microcircula-
Vessel Diameter, pm tion, red cell volume fraction will decrease in these
FIG. 3. In vivo data on vessel (tube) hematocrits in vessels of microvessels in which the surface roughness is significant
microcirculations. A : average vessel hematocrit (Hmicro) normalized by relative to vessel radius.
division by systemic hematocrit (Hsys) as a function of vessel diameter Difficulties with this concept arise from the postulate
and type. General trend is that Hmicro decreases as vessel diameter of the surface layer of plasma remaining stationary in
decreases, reaching values which are 40% of Hsys, which is a measure
of hematocrit of blood flowing into entire vascular bed. B: general
the face of a pressure differential that would cause flow
perception is that heterogeneity of vessel hematocrits increases as of an essentially Newtonian fluid, to say nothing of the
vessel size decreases. This is borne out by these in vivo data which very high velocity gradient generated between the sta-
show that coefficient of variation of vessel hematocrits for vessels of a tionary surface layer and a passing red cell.
given size and type increases as vessel diameter decreases. Dispropor-
tionate red cell distribution of successive arteriolar-type bifurcations
predicts such a behavior. Concept B
is clearly more than can be accounted for by error of The low capillary hematocrit could also result from
measurement, additional explanations are required. 3) repeated phase separation taking place at the successive
Capillary hematocrits have been reported to vary with microvascular bifurcations preceding the capillary net-
blood flow. Pharmacological and metabolic vasodilation work proper. This would cause variation of Hn between
elevates mean capillary hematocrit levels, whereas vaso- two daughter branches originating from a parent vessel
constriction lowers capillary hematocrit (34, 35). and would therefore lead to corresponding variations of
Two general avenues of thinking have been utilized to HT. Because of the nonuniform radial distribution of red
reconcile these findings that cannot be explained by the cells in a parent vessel, red cells and plasma are not
classical single-vessel Fahraeus effect alone. Both of evenly distributed at bifurcations, and in general, the
these accept Fahraeus’ original considerations as well as faster flowing branch receives the higher hematocrit.
their applicability to flow in the microcirculation. Al- However, the elevation of Hn in the faster branch is less
though concept A (13, l&34) is based on the assumption than the reduction of Hn in the slower, and the average
(supported by direct measurement) that the discharge of the two, just as the mean of the two ensuing Hrs, is
hematocrit (Hn) is the same in all microvessels and lower than that of the parent vessel (9, 48). Therefore,
therefore HT/Hn < 0.5 requires additional explanation, nonuniform distribution of flow tends to induce a differ-
concept B (9, 48) calls for variation of discharge hema- ence in the overall velocity of the red cells traversing the
tocrit within a microvessel network and maintains that microvessel network compared with that of the plasma.
H1014 ROBIN FAHRAEUS: 1888-1968

The mean Hn of the network will be less than the considerable interest in the context of inflammatory
hematocrit of the feeding and draining vessels, and mean reactions and postischemic reperfusion injury.
HT will be ~0.5 Hsys. This effect is a direct conceptual It turns out that Fahraeus’ original concepts, which
generalization of the original Fahraeus effect and has were mainly aimed at a better understanding of circula-
therefore been termed the network Fahraeus effect (39, tory hemodynamics, have evolved into complex exten-
49). Conceptually it replaces the higher flow velocities in sions of fundamental relevance for oxygen transport and
the centerstream of a single vessel by the higher flow its regulation on a local basis. Much less progress has
velocities in one of the daughter vessels of an arterial been made in the hemodynamic implications of his work;
bifurcation and assumes that red cells will preferably in particular, the role of red cell aggregation for the
follow these faster pathways because of phase separation resistance to blood flow and its distribution in microvas-
just as they follow the central streamlines in the single cular networks has still not been identified in vivo despite
vessel because of axial migration. some efforts. Although the pathophysiology of the car-
Evidence in support of this concept has been provided diovascular system may therefore not have gained so
by showing that the observed hematocrit distributions much from pursuit of Fahraeus’ ideas, circulatory phys-
can be explained on the basis of the combined network iology obviously has.
and single-vessel Fahraeus effects if the topological
structure of the microvascular system is known. Fur- Support from the Alexander von Humboldt-Stiftung under its
thermore, the high coefficient of variation observed in United States Senior Scientist Program (to G. R. Cokelet) and from
the Deutscher Akademischer Austauschdienst (to H. L. Goldsmith) is
those studies in which all rather than selected capillaries gratefully acknowledged.
of a given network were analyzed strongly supports this Address for reprint requests: H. L. Goldsmith, McGill University
concept (9, 39). Medical Clinic, Montreal General Hospital, 1650 Cedar Ave., Montreal,
The physiological implications for oxygen transport to Quebec H3G lA4, Canada.
the tissue are quite different in both concepts. Because Received 8 November 1988; accepted in final form 16 May 1989.
the product of discharge hematocrit and volume flow is
proportional to the oxygen delivery through a microves- REFERENCES
se1 and because the tube hematocrit in the capillaries
may influence the local rate of oxygen extraction (22), 1. BARBEE, J. H., AND G. R. COKELET. The Fahraeus effect. Micro-
vast. Res. 3: 6-16, 1971.
the presence or absence of supply heterogeneity as well 2. BAYLISS, L. E. Rheology of blood and lymph. In: Deformation and
as its possible control by flow-regulating mechanisms Flow in Biological Systems, edited by A. Frey-Wyssling. New York:
critically depends on the clarification of the above issues. Interscience, 1952, p. 355-418.
This is a problem of considerable general interest, since 3. BINGHAM, E. C., AND H. GREEN. Paint, a plastic material and not
a viscous liquid; the measurement of its mobility and yield value.
the exchange efficiency of heterogeneous networks is less Proc. Am. Sot. Test. Muter. 19: 640-664, 1919.
than that of homogeneous networks. Several studies have 4. BLIXT, A., M. BRAIDE, R. MYRHAGE, AND U. BAGGE. Vital micro-
shown beyond doubt that microvessel hematocrits in- scopic studies on the capillary distribution of leukocytes in rat
crease substantially with vasodilation and vice versa. If cremaster muscle. Int. J. Microcirc. Clin. Exp. 6: 273-286, 1987.
5. BRENNER, H. Hydrodynamic resistance of particles at small Rey-
this involves changes in discharge rather than only tube nolds numbers. In: Advances in Chemical Engineering, edited by
hematocrits, oxygen delivery to tissue will not be pro- T. B. Drew, J. W. Hoopes, and T. Vermuelen. New York: Academic,
portional to blood flow, since the variation of hematocrit 1966, vol. 6, p. 287-438.
will provide a large gain factor. 6. CHAN, P. C.-H., AND L. G. LEAL. The motion of a deformable drop
in a second order fluid. J. FZuid Mech. 92: 131-170,1979.
So far, the differences between experimental observa- 7. COKELET, G. R. Experimental determination of the average hem-
tions, as well as the resulting conceptual discrepancies atocrit of blood flowing in a vessel. Microvasc. Res. 7: 382-384,
pointed out, represent an unresolved yet physiologically 1974.
relevant problem. It should be noted that comparisons 8. COKELET, G. R. Macroscopic rheology and tube flow of human
blood. In: Microcirculution, edited by J. Grayson and W. Zingg.
between hematocrit distributions obtained experimen- New York: Plenum, 1976, vol. 1, p. 9-31.
tally in different tissues are complicated by the fact that 9. COKELET, G. R. Speculation on a cause of low vessel hematocrits
the detailed structure of the vessel network is often not in the microcirculation. MicrocircuZation 2: 1-18, 1982.
known and thus may lead to a bias in the procedure for 10. COPLEY, A. L., AND G. STAINSBY. Flow Properties of Blood and
Other Biological Systems. New York: Pergamon, 1960, p. 446.
collecting data. Because network structure is clearly im- 11. CORATTIYL, V., AND E. C. ECKSTEIN. Regional platelet concentra-
portant for the resulting dispersion of hematocrits (39), tion in blood flow through capillary tubes. Microvusc. Res. 32: 261-
a final resolution of the present discrepancy will have to 270,1986.
await more complete information from different tissues 12. COX, R. G., AND S. K. HSU. The lateral migration of solid particles
in a laminar flow near a plane wall. Int. J. Multiphase Flow 3: ZOl-
and different animal species. 222,1977.
In principle, such considerations, which are derived 13. DESJARDINS, C., AND B. R. DULING. Microvessel hematocrit: meas-
from Fahraeus’ original concepts, can also be applied to urement and implications for capillary oxygen transport. Am. J.
the differential distribution of other blood cells in the Physiol. 252 (Heart Circ. Physiol. 21): H494-H503, 1987.
14. DESJARDINS, C., AND B. R. DULING. Endothelial cell surface
vascular system. In the case of white blood cells for glycosaminoglycans are involved in the regulation of capillary
instance, uneven intranetwork distribution results from hematocrit (Abstract). FASEB J. 2: 945, 1988.
the intravessel radial distribution (4, 40), and this can 15. DULING, B. R., AND C. DESJARDINS. Capillary hematocrit-what
be affected by red cell aggregation at low flow rates. does it mean? News Physiol. Sci. 2: 66-69, 1987.
16. FWHRAEUS, R. The suspension-stability of the blood. Acta Med.
Therefore, a redistribution of white blood cell delivery to Stand. 55: l-228,1921.
subnetworks of the microvascular system could be 17. FWHRAEUS, R. Die Stromungsverhaltnisse und die Verteilung der
brought about by alteration of blood flow. This is of Blutzellen im Gefassystem. KZin. Wochenschr. 7: 100-106, 1928.
 ROBIN FAHRAEUS: 1888-1968 H1015
18. FAHRAEUS, R. The suspension stability of the blood. Physiol. Reu. of cat mesentery. Circ. Res. 43: 738-749, 1978.
9: 241-274,1929. 42. LIPOWSKY, H. H., S. USAMI, AND S. CHIEN. In vivo measurements
19. FAHRAEUS, R. The influence of the rouleau formation of the of “apparent viscosity” and microvessel hematocrit in the mesen-
erythrocytes on the rheology of the blood. Acta Med. Stand. 161: tery of the cat. Microvasc. Res. 19: 297-319, 1980.
M-165,1958. 43. LIPOWSKY, H. H., AND B. W. ZWEIFACH. Methods for the simul-
FWHRAEUS, R. Die Grundlagen der neueren Humoralpathologie. taneous measurement of pressure differentials and flow in single
Die fruhe Geschichte der Mikrozirkulation. Virchows Arch. Path. unbranched vessels of the microcirculation for rheological studies.
Anat. 333: 176-189,196O. Microvasc. Res. 14: 345-361, 1977.
FAHRAEUS, R., AND T. LINDQVIST. The viscosity of the blood in 44. MAUDE, A. D., AND R. L. WHITMORE. Theory of blood flow in
narrow capillary tubes. Am. J. Physiol. 96: 562-568, 1931. narrow glass tubes. J. AppZ. Physiol. 12: 105-113, 1958.
FEDERSPIEL, W., AND I. SARELIUS. An examination of the contri- 45. NOBIS, U., AND P. GAEHTGENS. Rheology of white blood cells
bution of red cell spacing to the uniformity of oxygen flux at the during flow of blood through capillary tubes. BibZ. Anat. 20: 2ll-
capillary wall. Microvasc. Res. 27: 273-285, 1984. 214,198l.
23. GAEHTGENS, P. Flow of blood through narrow capillaries: rheolog- 46. NOBIS, U., A. R. PRIES, AND P. GAEHTGENS. Rheological mecha-
ical mechanisms determining capillary hematocrit and apparent nisms contributing to WBC-migration. In: White Blood Cells: Mor-
viscosity. Biorheology 17: 183-189, 1980. phology and Rheology Related to Function, edited by U. Bagge, G.
24. GAEHTGENS, P., K. H. ALBRECHT, AND F. KREUTZ. Fahraeus effect V. R. Born, and P. Gaehtgens. The Hague: Nijhoff, 1982, p. 57-65.
and cell screening during the flow of blood. I. Effect of variation 47. PAPPENHEIMER, J. R. Contributions to microvascular research of
of flow rate. Biorheology 15: 147-154, 1978. Jean Leonard Marie Poiseuille. In: Handbook of Physiology. The
25. GOLDSMITH, H. L. Red cell motions and wall interactions in tube Cardiovascular System. Bethesda, MD: Am. Physiol. Sot., 1984,
flow. Federation Proc. 30: 1578-1588,197l. sect. 2, vol. IV, pt. 1, chapt. 1, p. l-10.
26. GOLDSMITH, H. L., AND G. R. COKELET. Decreased resistance in 48. POPEL, A. S. Effect of heterogeneity of capillary flow on the
the two-phase flow of blood through small tubes at low shear rates capillary hematocrit. Proc. Am. Sot. Mech. Eng. AppZ. Mech. Div.
(Abstract). Biorheology 23: 279, 1986. 32: 83-84, 1979.
GOLDSMITH, H. L., AND J. MARLOW. Flow behavior of erythro- 49. PRIES, A. R., K. LEY, AND P. GAEHTGENS. Generalization of the
cytes. II. Particle motions in concentrated suspensions of ghost Fahraeus principle for microvessel networks. Am. J. Physiol. 251
cells. J. CoZZoid Interface Sci. 71: 383-407, 1979. (Heart Circ. Physiol. 20): H1324-H1332, 1986.
28. GOLDSMITH, H. L., AND S. G. MASON. The microrheology of 50. REINKE, W., P. GAEHTGENS, AND P. C. JOHNSON. Blood viscosity
dispersions. In: Rheology: Theory & Applications, edited by F. R. in small tubes: effect of shear rate, aggregation and sedimentation.
Eirich. New York: Academic, 1967, vol. IV, p. 85-250. Am. J. Physiol. 253 (Heart Circ. Physiol. 22): H540-H547, 1987.
GOLDSMITH, H. L., AND S. SPAIN. Margination of leukocytes in 51. SARELIUS, I. H., AND B. R. DULING. Direct measurement of micro-
blood flow through small tubes. Microvasc. Res. 27: 204-222,1984. vessel hematocrit, red cell flux, velocity and transit time. Am. J.
HESS, W. R. Gehorcht das Blut dem allgemeinen Stromungsgesetz Physiol. 243 (Heart Circ. Physiol. 12): H1018-H1026, 1982.
der Flussigkeiten? Pfluegers Arch. 164: 603-644, 1915. 52. SCHKLAREWSKY, A. Ueber das Blut und die Suspension-flussigkei-
JOHNSON, P. C., J. BLASCHKE, K. S. BURTON, AND J. H. DIAL. ten. Pfluegers Arch. 1: 603-644, 1868.
Influence of flow variations on capillary hematocrit in mesentery. 53. SCOTT-BLAIR, G. W. The importance of the sigma phenomenon in
Am. J. Physiol. 221: 105-112, 1971. the study of the flow of blood. RheoZ. Acta 1: 123-126,1958.
32. KANZOW, G. A., A. R. PRIES, AND P. GAEHTGENS. Analysis of the 54. SCHMID-SCH~NBEIN, G. W., AND B. W. ZWEIFACH. RBC velocity
hematocrit distribution in the mesenteric microcirculation. Int. J. profiles in arterioles and venules of the rabbit omentum. Microuusc.
Microcirc. CZin. Exp. 1: 67-79, 1982. Res. 10: 153-164, 1975.
55. SEGRI?, S., AND A. SILBERBERG. Behaviour of macroscopic rigid
33. KARNIS, A., H. L. GOLDSMITH, AND S. G. MASON. The flow of
spheres in Poiseuille flow. Part 2. Experimental results and inter-
suspensions through tubes. V. Inertial effects. Can. J. Chem. Eng.
pretation. J. Fluid Mech. 14: 136-157, 1962.
44: 181-193,1966.
56. SHESHADRI, V., AND S. P. SUTERA. Concentration changes of
KLITZMAN, B., AND B. R. DULING. Microvascular hematocrit and suspensions of rigid spheres flowing through tubes. J. CoZZoid
red cell flow in resting and contracting striated muscle. Am. J. Interface Sci. 27: 101-110, 1968.
Physiol. 237 (Heart Circ. Physiol. 6): H481-H490, 1979. 57. SHESHADRI, V., AND S. P. SUTERA. Apparent viscosity of coarse,
35. KLITZMAN, B., AND P. C. JOHNSON. Capillary network geometry concentrated suspensions in tube flow. Trans. Sot. Rheol. 14: 35l-
and red cell distribution in hamster cremaster muscle. Am. J. 373,197o.
Physiol. 242 (Heart Circ. Physiol. 11): H211-H219, 1982. 58. TANGELDER, G. F., D. W. SLAAF, H. C. TEIRLINCK, R. ALEWIJNGE,
36. KNISELEY, M. H. Intravascular erythrocyte aggregation (blood AND R. S. RENEMAN. Localization within a thin optical section of
sludge). In: Handbook of Physiology. Circulation. Bethesda, MD: fluorescent platelets flowing in a microvessel. Microuusc. Res. 23:
Am. Physiol. Sot., 1965, sect. 2, chapt. 63, vol. III, p. 2249-2292. 214-230,1982.
37. KROGH, A. The Anatomy and Physiology of Capillaries (2nd ed.). 59. THOMA, R. Die Viskositat des Blutes und seine Stromung im
New Haven, CT: Yale Univ. Press, 1929. Arteriensystem. Deutsches Arch. KZin. Med. 99: 565-636, 1910.
38. LEAL, L. G. Particle motions in a viscous fluid. Annu. Reu. Fluid 60. THOMAS, H. W. The wall effect in capillary instruments: an
Mech. 12: 435-476,198O. improved analysis suitable for application to blood and other
39. LEVIN, M., B. DAWANT, AND A. S. POPEL. Effect of dispersion of particulate suspensions. BiorheoZogy 1: 45-56, 1962.
vessel diameters and lengths in stochastic networks. II. Modeling 61. VAND, V. Viscosity of solutions and suspensions. I. Theory. J.
of microvascular hematocrit distribution. Microvasc. Res. 31: 223- Phys. CoZZoid Chem. 52: 277-299, 1948.
234,1986. 62. VEJLENS, G. The distribution of leukocytes in the vascular system.
40. LEY, K., A. R. PRIES, AND P. GAEHTGENS. Preferential distribution Acta Path. Microbial. Stand. 33, Suppl.: 11-239, 1938.
of leukocytes in rat mesentery microvessel networks. Pfluegers 63. WHITTAKER, S. R. F., AND F. R. WINTON. The apparent viscosity
Arch. 412: 93-100,1988. of blood flowing in the isolated hindlimb of the dog, and its
41. LIPOWSKY, H. H., S. KOVALCHECK, AND B. W. ZWEIFACH. The variation with corpuscular concentration. J. Physiol. Lond. 78:
distribution of blood rheological narameters in the microcirculation 339-369,1933.