Clinical Eiomerhanics Vol. 11. No. 5, pp. ?

S3-25Y, 1996
Copynght 0 1996 Published by Elsevicr Science Limited. All rights rcscrved
Printed in Great Britain
0268-0033196 $15 .OO + 0.00

ELSEVIER

0268-0033(95)00068-2

Relationship between vertical ground

reaction force and speed during walking,
slow jogging, and running
T S Keller’, A M Weisberger*, J L Ray3, S S Hasan4, R G Shiavi4,
D M Spengler*
‘Department of Mechanical Engineering, University of Vermont, ‘Department of
Orthopaedics and Rehabilitation, 3Department of Mechanical Engineering, and
4Department of Biomedical Engineering, Vanderbilt University, USA

Abstract
Objective. To obtain descriptive information between vertical ground reaction force
(GRF)-time histories and gait speed, running style, and gender.
Design. GRF-time history measurements were obtained from male and female subjects
during walking, slow jogging, jogging and running on an indoor platform.
Background. Previous studies have established GRF descriptor variables for male subjects
running at speeds from 3 to 6 m s-‘, but very little descriptive data exists for slower or faster
running, nor have previous studies reported GRF descriptors separatelyforfemale subjects.
Methods. GRF-time histories were recorded for 13 male and 10 female recreational athletes
during walking and slow jogging at speeds between 1.5 and 3.0 m s-l, and running at
speeds between 3.5 and 6.0 m s -‘. Vertical GRF-time data for trials with speeds within
0.2 m s-’ of the prescribed speed were analysed to determine thrust maximum GRF (F,)
and loading rate (G,).
Results. In both male and female subjects, F, increased linearly during walking and running
from 1.2 BW to approximately 2.5 BW at 6.0 m s-‘, remaining constant during forward lean
sprinting at higher speeds. F, was linearly correlated to G,, the latter ranging from 8 to
30 BW s-’ over this speed range. Slow jogging was associated with a > 50% higher F, and
G, in comparison to walking or fast running.
Conclusions. Similar GRF descriptor data and velocity relationships were obtained for male
and female subjects. Impact forces were greatest when the subjects adopted a higher, less
fixed centre of gravity during slow jogging.

Relevance
These results suggest that vertical GRF norms can be established for male and female
subjects alike, and that slow or fast running with a lower, fixed centre of gravity decreases
impact forces. Copyright @ 1996 Published by Elsevier Science Ltd.

Key words: Gait, ground reaction force, thrust maximum, speed, running, biomechanics

C/in. Biomech. Vol. 11, No. 5, 253-259, 1996

Introduction occurred among dedicated distance runners logging

The popularity of recreational running has increased
dramatically over the past few years, as has the
incidence of overuse or repetitive loading injuries.
Clinical evidence suggests that workout intensity plays
a major role in the development of overuse injuries.
In a study *conducted by James and associates’ in-
volving 180 patients. 65% of the chronic injuries
Received: 25 January 1995; Accepted: 13 October 1995
Correspondence and reprint requests to: Tony S Keller
University of Vermont, Department of Mechanical Engineering,
119 Votey Building, Burlington, VT 05405-0156, USA
high mileage on a daily basis. Two-thirds of the chronic
injuries were attributed to high mileage, workout
intensity, running up hills and on hard surfaces, and/or
rapid change in training routine. Other researchers
have postulated that impact forces associated with
repeated loading are responsible for certain types of
overuse injuries of the musculoskeletal system’.“.
The notion that there may be a positive relationship
between impact force and overuse injuries during
running, together with the need to assess athlete
PhD,
performance, has prompted numerous experimental
studies of ground reaction force (GRF)-time histories
during the past 20 years4. Ground reaction force-time
 254 Ctin. Biomech. Vol. 11, No. 5, 1996
histories provide dcscriptil e information concerning
the magnitude. direction and point of application of the
impact fori~:: In general the vertical component of the
GRF domimites the impact force--time histor) in
comparison t/1 the other two components (backward-.
forward. me&al-lateral). and hence is the easiest to
quantify. The vertical GKI’ also shows the least vari-
abitit! between and within subjects’ i. Studies have
indicated that the descriptive data characterizing the
verlicitl GRE (loading rate. impact peak. relative
minimum. :hl u\t maximum. decay rate) are dependent
up0n numcroiis external factors such its suhjcct body
masi. loading rate. running speed. running stvle, area
tri !hc foot sround contact. :IS well as the mechanical
lm~p~rtic’s of’ the fool. shoe. ;Ind surface involved’~” “.
<ii thy nl:\!?v esterna! tactors which intluencc the
GRF. gait ++~ed has been the central focus of man)
- !(I
invcstigiltiony ’ i ’ .I’hcs1: studies have consistentI!
rtcttcd thilt. rhr: nla,gnitutie clf the vertical CiRE‘ increasec;
witli rni,rc;t3iti!; speed %>?‘cr thy range 111‘ spends
~~xrtmined. i>:lr h;~vc genr~rall\ been limited to a narro\\
tati~c 01 \\,;iliiing 01 runnirtg speeds ;md:‘or 3 small
1rumt~r <,I x?ibjcctb. In :]I! c>ffort TV)establish reference
stanthrrdu l’<)r C;RF dat,r as il function of running speed.
Munr~) CI ~1.. .:oliccted GRF data from 20 adult males
a1 \ptc’(fs rzmging from i-5 m 5 ’ tit the SQWdS
~x;m~~nc~l. th majorirt of the subjects in this stud!
\+tcrc rC;ir--t(>~!I 5trikcrs LL~OSCGKF-time history wan,
~h;ir;+~t~~r~~c:c!t)\ ;I,] iriiti;jl sharp peak (impact Inax-
mum) i’olic~t:~l hi ;I sccc,nd pc;ik at mid-stance (thrust
m;t*imilJn! i hc( nored that rhr impact maximum
irrcrc.;\\etl i airagut i .i-Iold) in it linear manner from
j CL%~lvwcigh! (l3W) t(i 2.S HW over this range 01
>pcctl~ - ir,irc;i5C:s iI, the thrust maximum and the
;~~,c~-:Q~c \trr?ir~ri C;KF cxcrtctl throughout the stance
phahz howc.\ J:- ~c’ce i+s remarkable ( I, I -fold and
1h..I“.11,ldi ;tbL’i ?hcxe running speeds. and exhibited ;I
??lOrL’ 11011. iir!t.;tr relationuhlp kvith i;peed (increasing less
wit!? ;noi-c;rslll,z3 running speed). Nipg et al: found
SilTiililj ti’?,ittii 1~11 14 rllalc\ running at speeds ranging
r”riK: .; f. iJ’l Whili~ these results establish useful
\tnndnr&. !%:I.(; KI-‘de>cripi;pr \,:lriables for nialc runners.
the; do noi l>r(!l J&Z descriptive data for women. nor do
thcW rcsuii~ ;lr pre\%~u\ IitCraturc consider running at
ifXW.l$ IC\S t!-c,ii\ : 111 \ ’ or grcatcl- that) f, 111x ‘,
‘Z%C,i>f>jt’cti\‘* of thE> <~uJy W;IS LO re-examine the
relationship between 111~ vertical CiKF and speed
tzncompassmg ;i wide r;\nge of physiological running
speeds. In particular WC‘wished to answer the following
qut,stions: (1 ) docc the vertical GKI~ increase in a lineal
manner at running speeds greater than S m s ‘?. (7)
does gender intluencc the vertical GKF‘.‘. and (3) what
ei’f~ct does sIo\~ .joggmg as opposed to fast walking
have on the vcrtlcal GKF’.’
Age iyears) 25.2 (SD 4.31 28.4 (SD 5.4)
Height (cm1 178.4 (SD 7.0) 168.3 (SD 7.0)
Mass(kg} 75.6 (SD 12.0) 57.6 (SD 5.81
Methods
Twenty-three subjects (13 males and 10 females) were
used in this study (Table 1). All subjects were recre-
ational athletes who participated on a regular basis in a
variety of activities including: basketball. squash, cyc-
ling, soccer. racquetball, distance running. volleyball,
tennis, weight lifting, triathlons. and other sports. All
were within the range of normal weight for their height.
The majority of subjects surveyed indicated that they run
at average speeds of X-c) minutes/mile for females and
7- S minutes/mile for males. ‘I’hese speeds correspond
approximately to 3.0-3.3 m s ’ and 3.4--3.X m s-’ for
females and males respectively. During the tests the
male and female subjects wore Nike Aircraft running
shoes with identical soles and cushioning.
Subjects were asked to walk. jog, and run over a
11-m running platform. ‘This arrangement provided 6 m
for the subjects to accelerate :md decelerate. Football
dummie\ were placed at the end of the runway as ;I
buffer during deceleration and subjects were en-
couraged to USC them at higher speeds. A A-channel
force pl;itform (,Model OR&3, Advanced Mechanical
Technology. Inc.. Newton. MA). with a natural
frequency 01’401) Hz. was located flush in the centrc of
the platform and was directly connected to a IWP I 1123
computer. A I:-bit A0 converter was used to sample
the GKF --time history data at 3% samples per second.
l‘his SilJllpling frequency was based upon Nigg’s’”
reconimendution that the appropriate frequency for
data acquisition should be at least five times the
maximum frequency content of the analysed signal.
During running up to 6 m s ’ the frequency content of
GRI; -time histories is not more than 3) lIzI-‘.
The force platform surface measured 50X mm x
357 mm and was outlined with a bright yellow. ‘r-mm
wide I;I~c’. Subjects were allowed as many practice
trials a\ needed to achieve acceptable foot contact
(within the bounds of the force platform minus 25 mm
on each side) and were given ;i rest of at least I min
between speed trials. A line of 20 1X1) lights. spaced
0.5 m apart and set to blink in sequence at the desired
specda. \tas placed OJI the margin of’ the running stage
in order to guide the subject. Subjects were required to
contact the force platform using the same foot (right or
I&t), because the data acquisition program required
consistent foot usage throughout. Data from both right
and left foot strikers were combined into distinct speed
categories according to the method of Munro and
associates’. An KGB video camera (Hitachi, model
KP-C’IOSA. Hitachi Denshi Ltd. Japan) and a h-head
videocassette recorder (I litachi. model VT-33OA,
Hitachi I.td. Japan) were used to film and record (at 30
frames per second) the foot-strike pattern and contact
angle during each trial. The camera was located
adjacent to the force platform facing either the medial
or lateral aspect of the foot, depending upon whether a
left or right foot strike occurred. respectively.
A minimum of four walking speeds (1.5. 2.0. 3.5.
3.0 m s .‘) and four running speeds (3.5. 4.0. 5.0,

6.0 m s-‘) were measured for males. Female subjects
were measured at the same walking and running speed
intervals up to 4.0 m s-l, after which the speed inter-
vals increased at 0.5 m SC’ intervals up to the subjects’
maximum speed. In order to obtain GRF-time
histories at the subjects’ maximum speed, male and
female subjects were encouraged to run as fast as
possible above 6.0 m s-’ and 5.0 m SC’ respectively. In
a subset of 12 subjects (6 males, 6 females), the subjects
were asked to slow jog at speeds of 1.5, 2.0, 2.5, and
3.0 m s-‘. Slow jogging was distinguished from walk-
ing by the absence of a double support phase. Walking,
slow jogging, and running speeds were measured by
two photoelectric cells located 1 m from the centre of
the force platform, and mounted so that the photo-
electric cells were triggered by the subjects’ waist. Up to
10 trials at each speed were recorded, and only trials in
which there was good foot contact within the perimeter of
the tape, a steady stride, and speeds within ?Z0.2 m SC’
of the prescribed speed were analysed.
The six-channel GRF-time history data was later
processed on a PDP 11173computer using data analysis
methods described previously”. The main variables
reported in this paper are the vertical thrust maximum
force (F,), vertical thrust maximum loading rate (G,),
and speed (v). In order to precisely determine the
magnitude and time duration of the vertical thrust
maximum force, a four-point interleave filter (3 db cut-
off = 15 Hz) was used to smooth the 256 samples/
second GRF-time history data. This smoothing
process produced data records containing 65 samples/s
(256/4 + 1). Thrust maximum loading rates were calcu-
lated by dividing F, by the time interval between initial
foot contact and the occurrence of the vertical thrust
maximum force. In accordance with Munro et al.‘, F,
and G, were normalized to the subject’s bodyweight
(BW). Vertical impact peak forces were not determined
from the GRF-time histories, since these short duration
peaks were attenuated by of the smoothing scheme used
to process the data. The smoothing scheme, however,
produced only a small reduction (about 2-5%) in the
thrust force values at the highest speeds.
Foot-contact patterns for each trial of each subject
were quantified by examining digitized images obtained
from the videotape recordings. Originally we had
intended to determine both the foot-strike index (rear-
foot, mid-foot, fore-foot)-i and the contact angle from
the video recordings of the foot-strike patterns. Both
are important parameters which are required for
dynamic analysis and modelling of rigid body motion of
the lower extremities. Subsequent analysis of digitized
images of each foot strike, however, indicated that
while this procedure was adequate for determining the
foot-strike index, we could not obtain accurate contact
angle measurements above 3 m s-’ with the frame rate
used (30 Hz). This paper, therefore, presents only the
former. It should be noted that one can perform centre
of pressure measurements to determine contact
patterns2,‘, but such measurements cannot be used to
compute contact angle measurements.
Keller et al: Ground reaction force and gait speed 255
Percent of Total ReDetitions
100
80
60
40
20
0
1 1.5 2 2.5 3 3.5 4 5 6 6.5 7
Speed (meters/second)
Forefoot E2 Midfoot 0 Reatfoot
Figure 1. Foot strike indices (rear-foot, mid-foot, fore-foot) versus speed
for all subjects. indices are depicted in terms of the percentage of total
walking and running repetitions (n = 8791. Foot strike patterns change
from predominantly rear-foot to predominantly mid-foot at 6 m 5-l.
Means and standard deviations (SD) of the descriptive
variables were obtained at each of the fixed walking,
slow jogging, and running speeds. Linear regression
models were also applied to the force-velocity data,
and R2 values and levels of significance were calculated
for the regression equations. An analysis of covariance
(ANCOVA, equality of slopes) was also performed to
determine if the regression models were significantly
different for male and female subjects. GRF descriptor
variable differences between men and women, at differ-
ent speeds, and between slow jogging and walking were
assessedusing a one-way analysis of variance (ANOVA).
Results
Analysis of the foot strike indices indicated that the
majority of subjects were rear-foot strikers at speeds
less than 5 m SC’ (Figure 1). At speeds above 3 m SC’
there was an increasing frequency of mid-foot and fore-
foot strikes. Eighty-six percent of the subjects were
mid-foot or fore-foot strikers at 6.0 m SC’. Eight
females achieved speeds of 5 m s-’ and two completed
five trials at 6 m s- ‘. All males achieved speeds of
6 m s-l and four completed four or more trials at
7 m s-‘. One male subject completed three trials at a
speed of 8 m SC’ using a rear-foot strike pattern. Many
subjects increased their stride length and assumed a
more crouched, forward leaning posture during their
high-speed running trials.
The vertical GRF-time histories exhibited a double
peak during walking and running below speeds of 2.5-
3.0 m SC’ (Figure 2). At these speeds the thrust maxi-
mum force was generally the first peak recorded and
occurred between 15 and 25% of the total stance time.
At higher running speeds, the GRF-time histories
consisted of a single peak (thrust maximum) located at
about 40-50% of the total stance time. The mean
values for F, ranged from 1.15 BW at 1.5 m SC’ to
2.54 BW at 4.5 m s-’ for females, and from 1.23 BW at
1.5 m SC’ to 2.46 BW at 5 m s-l for males (Table 2).
The average loading rate increased from 7.77 to
30.0 BW SC’ and 8.20 to 29.1 BW SC’ in the speed
range 1.5-6.0 m SC’ for the female and male subjects
respectively.
256 C/in. Biomech. Vol. 11, No. 5, 1996

250

Time (% Stance) Time (% Stance)

Figure 2. Anterior-posterior vertical GRF-time histories patterns as functions of running speed. Time histories shown were smoothed using a 4-point
interleave filter and normalized as a percentage of the total stance time according to the method of Hasan et al. 14, Impact peaks were not present in the
vertical GRF- time histories because of the smoothing procedure used to post-process the data. (a) Female subject (7) for speeds of 1.5-6.0 m S- ‘.
Transition from double to single vertical force peak occurs at 2.5 m s ‘. (b) Male subject (9) for speeds of 1.5-7.0 m sm.‘. Transition from double to single
veltical force peak occurs a? 3.0 m q ’

‘11~ icrtlcai thrust maximum force increased in a
lineal- manner with increasing speed up to abot~t
1 ,? !I-, c ’ for both males and females (,Figure 3). V’ari-
:~l~ms in 2;, wt:re greatest in the speed transition region
l>riM;<$<:n lvalking and running (e.g. 2.5.-3.0 m s ‘) at
which point some subJccts walked and some jogged.
ii; 3.5 nl :. the male and female subjects were run-
ning at 53,11”:1(SD 5.2) and 67.5(X, (SD 6.1) of their
maximum speed respectively. Linear regression equa-
rions and the coefficient of determination (R’) for F,
(RN’) vc’rsux speed (walking and running gaits) in the
range (11I 5 ~TI4- I c: i’ -C 3.5 m s -’ were:
Table 2. Summary of vertical GRF variables (mean values)
Males (n = 291) F, = 0.598 L’ + 0.249.
R” = 0.65 (PCO.001)
Females (n = 240) F, = 0.634 v + 0.159,
R” = 0.66 (PCO.001)
where n is the number of trials. Incremental changes in
F, were statistically significant (ANOVA, P-cO.05) up to
3.5 m s-l for both male and female subjects. At speeds
greater than about 3.5 m s-l there were no significant
increases in F, for either group of subjects. In the male
subjects there was a slight decrease in F, at the highest
speeds. particularly for the subject who ran up to
grouped by running speed and sex

Females Males
__--
Thrust Thrust
max. farce Loading rate Speed max. force Loading rate
(F,, BWJ fG, BW s ‘i (lt 0.2 m s ‘) Fz, SW! (G, BW s-‘I
~___--
1.5 (r: 501 1.15 (0.10) 7.77 (1.781 1.5 (n = 65) 1.23” lO.10) 8.20 (1.84)
2.0 in 50; 1.36 (0.18) 11.5 (2.36) 2.0 (n = 641 1.42’ (0.14) 11 .o (2.29)
2.5 /n .= 49) ‘I .73 (0.43) 14.6 (3.71) 2.5 h = 65) 1.62 (0.24) 14.6 (2.46)
3.0 (r> = 50) 2.11 (0.46) 16.9 (3.97) 3.0 In = 61) 2.10 (0.50) 16.0 (3.30)
3.5 (a -= 41 J 2.36 (0.25) 19.1 (3.82) 3.5 (n = 37) 2.45 (0.28) 18.32 (3.36)
4.0 (n 46) 2.33 (0.32) 19.6 (4.65) 4.0 In = 58) 2.35 (0.48) 18.9 (4.85)
4.5 (n IO! 2.54 (0.27) 23.7 (4.91)
5.0 in = 38; 2.28 (0.32) 22.3 (4.61) 5.0 (n = 60) 2.46* (0.33) 22.8 (4.51)
5.5inx 101 2.13 (0.32) 22.5 (6.87)
6.C (P IO! 2 45 !0.13) 30.0 (2.63) 6.0 (n = 67) 2.38 (0.28) 29.1 (15.2)
6.58 (n ; 26) 2.34 (0.23) 37.8 (29.3)
7.0% (n = 17) 2.29 (0.19) 36.5 (22.5)
8.0% (n = 3) 1.89 (0.49) 58.5 (37.6)
so m parentheses. n - number of trials
f Significant difference ~ANOVA, P<O.O5) compared to females.
*iAopruximate running speed across force platform since subjects were accelerating between 4 and 6 mare speed measurement interval
 Keller et al: Ground reaction force and gait speed 257

3- 3.57

2.8 -
$3.0:
IL” 0
iz2., : g 2.5- 0
-1; I b

g22.4 :
JI I
8
U
E
3 2.2-
E : I
E 2.0-
.m c
i
- 1 3
2 2-
2 1.5-
I
2 HI-
t; - 0
c ;
3 1.6-
2
0 Females - walk, run f 1.0-
.u
- / 0 Male
5 0 Males - walk, ran
> 1.4- 0 Female

1.2-

0.01 I I 81, i I I I I I I II ,I -0 1 I
0 10 20 30 40
Thrust Maximum Loading Rate (BWs”)
Figure 3. Comparison of male (open squares) and female (open circles) Figure 4. Vertical thrust maximum force (F,) versus thrust maximum
vertical thrust maximum force versus speed (1.5-6.0 m ss’). Mean and loading rate (G,) for male (open squares) and female (open circles) during
standarddeviationsareshown.Bestfitlineforcombinedmaleandfemale walking and running. Both male and female subjects exhibited a similar
subjects is also shown for speeds up to 3.5 m s-’ (see text for linear positive linear relationship between F, and G,. Best fit line for combined
regression equation). Differences between male and female subjects male and female subjects is shown in the range 2.9 < G, < 26 BW s-’
were significant (ANOVA, P<O.O5) at speeds of 1.5,2.0, and 5.0 m s- ‘. (see text for linear regression).

8 m s-l (F, = 1.89, SD 0.49 m s-‘. Changes in G, were

(Table 3). Differences in F, and G, for slow jogging
also linear with regards to speed throughout the range
versus .walking were statistically significant (ANOVA,
of walking and running speedsexamined. However, the
P<O.OOl) at speeds ranging from 1.5-2.5 m s-r in
relationship between F, and G, was most linear only up
female subjects and 1.5-3.0 m s-l in male subjects.
to about 26 BW s-l, after which F, remained relatively
Females exhibited a smaller difference in forces
constant (Figure 4). The following linear regression
between slow jogging and walking than males. Both
equation and coefficient of determination (R*) was
groups indicated that walking was preferable to slow
obtained for F, (BW) versus G, during walking and
jogging or ‘slogging’.
running:
Males (n = 436) F, = O.O89G, + 0.520, Discussion
R* = 0.79 (P<O.OOl)
In this study, GRF-time histories and foot-strike
Females (n = 356) F, = O.O90G, + 0.482,
indices were analysed for 23 young male and female
R2 = 0.77 (P<O.OOl)
recreational athletes during walking, slow jogging, and
where G, < 26 BW SK’. running on a force platform. Normative data for
An ANOVA indicated that the difference in F, (BW) vertical GRF descriptor variables (thrust maximum,
between male and female subjects was significant for average loading rate) were presented and relationships
the following gait speeds: 1.5, 2.0 and 5.0 m s-l, but between the GRF descriptors and speed were studied.
these differences were small (<8%). There were no The notion that altered running gait (slow jogging
significant differences in G, (BW s-l) between the male versus walking) may influence the GRF-time histories
and female subjects at any of the speeds examined. was also examined. In order to establish normative
An ANCOVA indicated that there were no significant GRF data, a relatively large number of subjects
differences between the force-velocity and loading wearing shoes with identical soles and cushioning was
rate-velocity linear regression equations (equality of studied. Over 1100 GRF-time histories and foot
slopes) obtained for the male and female groups. contact patterns were collected and analysed for
Consequently the data for male and female subjects walking, slow jogging, and running at speeds ranging
was combined, yielding the following linear regression from 1.5 m s-l -8.0 m s-l. Despite limitations in the
relationships: runway length, most of the male and two of the female
recreational athletes examined in this study were able
Males + females (n = 531) F, = 0.614 v + 0.208,
to achieve constant running speeds up to 6 m s-‘.
R* = 0.65 (P<O.OOl)
Males + females (n = 753) F, = O.O89G, + 0.503, The magnitudes of the vertical thrust maximum
forces obtained in this study for walking (less than
R* = 0.78 (P<O.OOl)
about 2.5-3.0 m s-‘) and running (greater than 2.5-
where 1.5 < v < 3.5 m s-’ and 2.9 < G, < 26 BW s-l. 3.0 m s-l) compare favourably with previously pub-
Thrust maximum forces and loading rates were as lished results*,3,S,9,“,‘2~IS-*O
. Results from these
much as 62 and 65% greater, respectively, during slow studies are summarized graphically in Figure 5 for
jogging than during walking at the same speed comparison to the present study. Examination of
258 C/in. Biomech. Vol. 11, No. 5, 1996
Table
jogging
3. Retative difference
and walking gaits
1%) in vertical
..___
Females
. . ---_-
Speed F, I;.
tl?.2ms ’
_ _-
15 49.0* 45.1
2.0 39.4” 13.5*
2.5 18.7* 0.3
3.a 5.6 1.3
*SwmiRcanr differrncr !nNovn , PcO.001) between
i-igurc 5 indicates that walking is associated with
vertical thrust maximum forces between 1.1 and
1.s 13w. The range of thrust maximum forces
measured during the fast running speeds most
commonly reported in the literature (4-4.5 m s ‘) are
about double that of walking gait values. which is also
consistent with the present study.
I)nly two prior studies have performed comprc-
hens& GRF--time history studies over a range of run-
ning speeds’,i. For speeds of 3-S m se-‘, Munro and
associat& reported vertical thrust maximum force
values ranging from 2.51 BW (SKI 0.21) to 2.83 BW
(SP 0 I?‘\. The majority of the subjects examined by
Munro and associates were rear-foot strikers. Nigg
ct at.” reported that vertical thrust maximum force
values ranging from 1.86 kN (SU 0.17) to 2.26 kN (XI
0.42) or approximately 2.6-3.2 BW (estimated using
mean BW reported) are produced during running at
1% h m 3. ’ Nigg and associates reported that the foot
contact pattern changed from a rear-foot strike to a
mid-foot strike pattern at the highest speeds. Their data
also suggest that the vertical thrust maximum forces
wcrc more variable at the highest speed, which was
speculated to be the result of’ variations in foot contact
patterns. WC observed a similar change in foot strike
pattern. but a more consistent force variability with
increasing running speed. Both of the aforementioned
studies’.’ indicated that the increase in the vertical
thrust maximum force during running was linear with
increasing speed in the range 3-6 m s ‘. We also noted
a Linear increase in the vertical thrust maximum force
with increasing speed, but the relationship between the
vertical thrust maximum force and speed in our study
was linear only up to 4.0 m se ’ or about SO-60% of
maximum running speed of the subjects. Above this
speed the vertical thrust maximum force remained
relatively constant.
Noteworthy. therefore. was the finding that the
vertical thrust maximum forces during running at
S--h m s i were longer in magnitude (about lo-20%)
than that of Nigg et al.’ and Munro et al.i An explana-
tion for this finding may be the fact that the subjects in
the present study adopted a forward leaning running
posture at higher speeds. particularly during their
tastcst running trials. A forward leaning running style
lowers the centre of gravity of the subjects. and reduces
the downward velocity of the head, arms and trunk at
touchdown, thereby reducing the magnitude of the
CRF in comparison to running with more upright
postures. Bobbert et al.” noted that a ‘groucho’ style
GRF descriptors between slow running (running while keeping the centre of gravity
low) at 3.6-4.2 m s-l produced GRF forces which
__-._____-__
Males were about 25% lower than normal heel-toe running
FJ G, at the same speeds. Our findings associated with
________ -.. forward leaning during running are consistent with this.
62.2X 65.4* An interesting finding in this study was that slow
49.6* 36.0” jogging or ‘slogging’ produced forces that were
42.5* 11.6’
l&6* 15.4x significantly greater than those during walking.
slow jog and walk.
Differences between walking and ‘slogging’ were
greatest in the male subjects, who exhibited vertical
thrust maximum forces at 2.5-3.0 m s-l which were
comparable to F, values during running at speeds
between 3 and 8 m s I. The greater ‘slogging’ versus
walking differences for the male subjects may be due to
the fact that most females are already jogging at a
speed of 3.0 m s ‘. whereas the majority of the male
subjects chose to walk at 3.0 m s -I. Walking versus
running preferences reflect differences in stature
among the subjects; males tended to be taller and had a
larger stride length. One explanation for the observed
increased vertical thrust maximum forces associated
with ‘slogging’ is the fact that both the male and female
subjects adopted a higher centre of gravity in order to
slow jog, which increases the downward velocity of the
head, arms and trunk, and therefore increases the
magnitude of the GRF. Although ‘slogging’ at walking
speeds was a less natural gait, none of the subjects had
any problems with this style of running. ‘Slogging’,
however, produced a more bouncy and jarring style of
running. Given the choice between ‘slogging’ and
walking at the lowest speeds, therefore, all subjects
350,
2 A
$
-
300
1
d -I
b 150
2
Symbols _ Literature
5 - - Present Study
2
”
100
50(,~,,,1,,,,,,,,,,,,,,,,,,,,,,
0 1 2 3 4 5 6 7
Speed Ims-‘)
Figure 5. Comparison of vertical thrust maximum force (F&speed data
from the literature with the results of this study (linear regression line for
male t female subjects). In contrast to that observed in previous studies
(symbols in plot), F, values obtained during running did not increase
significantly above speeds of 3.5 m s ‘. This was hypothesized to be the
result of the lower centre of gravity associated with the fotward leaning
running posture adopted by most subjects. Symbol key: A --- Alexander
and Jayes” (7 male subjects, age range 7 23-58 years); - Munro et
al.’ (20 male subjects, mean age = 25.4 years); A 7 Nigg et al.’ (14 male
subjects, mean mass = 73 kg, F, data normalized with respect to mean
mass of subjects); * = Bates et al. ll; c = Cavanagh and Lafortune3 (10
male and 7 female subjects, mean age 24 years); + 7 Frederick et al.” (6
male and 3 female subjects); :> = Dickinson et al.16 (6 male subjects,
mean age 7 26.3 years); H = Hamill et al.“; c: - Roy”.

indicated that they would prefer to walk. Based upon
the fact that ‘slogging’ produced a significant increase in
vertical GRFs, this type of running style should be
avoided, since increased forces are highly undesirable
from the point of view of injury prevention. In this
regard, a running style with a lower centre of gravity
would appear to be optimal for minimizing GRFs.
To our knowledge this study represents the first
comprehensive study in which vertical GRF descriptors
are reported separately for male and female subjects.
Both the men and women subjects examined in this
study were young recreational athletes, active in a
number of different sports. While statistically signifi-
cant differences in F, values were noted between males
and females at some walking and running speeds, these
differences were generally unremarkable. The small
differences noted may reflect the fact that the female
subjects changed from walking to jogging at a lower
speed, and also had a lower maximum running speed.
More noteworthy was our finding that similar F,-
velocity and F,-G, relationships were obtained for
both sexes. Our results suggest that normative vertical
GRF descriptor data and velocity relationships can be
applied to both male and female subjects alike.
Additional work is needed to ascertain whether or not
these observations apply to other GRF descriptors such
as vertical impact force and impact loading rate.
Conclusion
Knowledge of the relationship between gait speed and
vertical ground reaction force is important for develop-
ing models of musculoskeletal adaptation to altered
activity, improvement of our understanding of the
aetiology and treatment of injuries associated with
running, and assessmentof running performance. The
results of this study of 13 men and 10 women recre-
ational athletes indicated that the vertical ground
reaction forces increased linearly with gait speed up to
about 60% of the subjects’ maximum speed. At higher
speeds, vertical forces remained constant at approxi-
mately 2.5 times body weight. This finding is different
from that reported previously for running at speeds of
3-6 m SC’, and is hypothesized to be the result of a
lower centre of gravity associated with the forward
leaning running style adopted by the subjects in order
to achieve fast running speeds on a relatively short
runway. Male and female subjects had similar values
for the vertical GRF descriptors at all the speeds
examined, and linear regression relationships between
vertical thrust maximum force and velocity were similar
for both groups of subjects. These results suggest,
therefore, that vertical GRF norms can be established
for males and females alike. Moreover, slow jogging or
‘slogging’, characterized by a higher centre of gravity
and more bouncy running style, produced vertical
forces as much as 1.6 times greater than normal walking
at the same speed or running at higher speeds. Running
style, therefore, appears to be a particularly important
determinant of vertical GRF-time history descriptor
variables.
Keller et al: Ground reaction force and gait speed 259
Acknowledgements
The authors would like to thank James P Bohan Jr,
Melvyn A Harrington Jr and Mike Stanfill for their
technical assistanceduring the study. We also thank Eko
Moeljanto for his assistance in computer programming.
References
1 James SL, BatesBT, Osternig LR. Injuries to runners.
Am J Sports Med 1978; 6: 40-50
2 Nigg BM, Bahlsen HA, Luethi SM, Stokes S. The
influence of running velocity and midsole hardness on
external impact forces in heel-toe running. J Biomech
1987; 20: 951-9
3 Cavanagh PR, Lafortune MA. Ground reaction forces in
distance running. J Biomech 1980; 13: 397-406
4 Miller DI. Ground reaction forces in distance running. In:
Cavanagh PR, ed. Biomechanics of Distance Running.
Human Kinetics, Champaign, IL, 1990; 203-24
5 Munro CF, Miller DI, Fuglevand AJ. Ground reaction
forces in running: a reexamination. J Biomech 1987; 20:
147-55
6 Nigg BM. External force measurements with sport shoes
and playing surfaces. In: Nigg BM, Kerr BA, eds.
International Symposium on Biomechanical Aspects of
Sport Shoes and Playing Surfaces. Univ. of Calgary Press,
Calgary, 1983; 11-23
7 Nigg BM, Denoth J, Neukomm PA. Quantifying the load
on the human body: problems and some possible
solutions. In: Morecki A, Fidelus K, Kedzior K, Wit A,
eds. Biomechanics VII-B. University Park Press,
Baltimore, 1981; 88-99
8 Radin EL, Orr RB, Kelman JL et al. Effect of prolonged
walking on concrete on the knees of sheep. J Biomech
1982; 15: 487-92
9 Bobbert MM, Schamhardt HC, Nigg BM. Calculation of
vertical ground reaction force estimates during running
from positional data. J Biomech 1991; 24: 1095- 105
10 Frederick EC, Hagy JL, Mann RA. The prediction of
vertical impact force during running. J Biomech 1981;
14: 498
11 Bates BT, Osternig LR, Sawhill JA, James SL. An
assessmentof subject variability, subject-shoe
interaction, and the evaluation of running shoes using
ground reaction force data. J Biomech 1983; 16: 181-91
12 Alexander R McN, Jayes AS. Fourier analysis of forces
exerted in walking and running. J Biomech 1980; 13:
383-90
13 Nigg BM. Experimental techniques used in running shoe
research. In: Nigg BM, ed. Biomechanics of Running
Shoes. Human Kinetics, Champaign, IL, 1986; 27-61
14 Hasan SS, Edmondstone MA, Limbird et al. Reaction
force patterns of injured and uninjured knees during
walking and pivoting. J Electromyogr Kinesiol1991;
1: 218-28
15 JacobsNA, Skorecki J, Charnley J. Analysis of the
vertical component of force in normal and pathological
gait. J Biomech 1972; 5: 11-34
16 Dickinson JA, Cook SD, Leinhardt TM. The
measurement of shock waves following heel strike while
running. J Biomech 1985; 18: 415-22
17 Hamill J, Bates BT, Knutzen KM, Sawhill JA. Variations
in ground reaction force parameters at different running
speeds.Hum Mvmt Sci 1983; 2: 47-56
18 Roy B. Caracteristiques biomechaniques de la course
d’endurance. Can J Appl Sci 1982; 7: 104-15
19 Frederick EC, Hagy JL, Mann RA. The prediction of
vertical impact force during running. J Biomech 1981;
14: 498
20 Hamill J, Bates BT, Knutzen KM. Ground reaction force
symmetry during walking and running. Res Q Exert Sport
1984; 55: 289-93