Downloaded from SAE International by University of New South Wales, Wednesday, September 19, 2018

740082

Analytical Studies of the

Head Injury Criterion (HIC)

Clifford C. Chou
Biomechanics Research Center, Wayne State Univ.

Gerald W . Nyquist
Research Laboratories, General Motors Corp.

T H E H E A D I N J U R Y C R I T E R I O N (HIC) is an analytical tool

currently recognized by the U. S. Department of Transporta-
tion, National Highway Traffic Safety Administration
( N H T S A ) , for determining if a blow to the head exceeds the
m a x i m u m tolerable severity, or injury threshold. Use of the
HIC is currently required in Federal Motor Vehicle Safety and the limits of integration tbegin and tend are the times at
Standard 208 ( F M V S S 208). Application of the criterion re- the onset and end of the impact, respectively. From Eq. 1 it
quires a knowledge of the head acceleration-time profile. Spe- is apparent that the acceleration is weighted by the exponent
cifically, the head c.g. linear acceleration vector magnitude 2.5, and therefore high accelerations for short time durations
time history, for the duration of the impact, is required. will contribute more to the integral than low accelerations for
The HIC evolved from a weighted impulse criterion known extended time durations.
as the Gadd Severity Index (GSI) (1),* which was previously A GSI value of 1000, calculated using the anterior-posterior
used in F M V S S 208 for evaluating head impacts, and is cur- (AP) component of acceleration, was originally recommended
rently required for chest impact evaluation. The GSI, and as the injury threshold criterion for frontal head impact (see
therefore the HIC, was developed to enable a methodical Ref. 1). W h e n N H T S A adopted the GSI, they redefined a(t)
means of comparing head impacts to biomechanical tolerance as the three-dimensional resultant vector magnitude rather
data in the literature (primarily the Wayne Tolerance Curve than the A P component, and maintained the 1000 injury
(2)). threshold value.
The GSI is defined by the relation Recently a n e w criterion, the HIC (3), has evolved. Mathe-
matically, the HIC is defined by the expression

*Numbers in parentheses designate References at end of where a(t) = resultant acceleration magnitude at head c.g.
paper. (g units) and t1 and t 2 are the two points in time during the

ABSTRACT
alysis is carried out for any arbitrary acceleration profile a(t).
The Head Injury Criterion (HIC) in F M V S S 208 for evalu- It is found that maximization requires that a(t1) = a(t2). Also,
ating the potential head injury requires maximization of a
for the unique values of t1 and t 2 that maximize HIC, the
mathematical expression, involving the time-average accelera-
tion, by varying the limits (t1,t2) of the time interval over average acceleration between t1 and t 2 is 5/3 times the acceler-
ation at t1 or t2. Illustrative examples are provided by apply-
which the average is calculated.
This paper describes the HIC behavior through the analysis ing this condition to simple pulses. Numerical results are pre-
of a function of two independent variables t1 and t2. The an- sented in tables and graphs.
398
 Downloaded from SAE International by University of New South Wales, Wednesday, September 19, 2018

impact which maximize HIC. Time is measured in seconds. sions. A discussion of the analytical results is included, and
The term in Eq. 2 raised to the power 2.5 is simply the average numerical results are presented in tables and graphs.
acceleration between t1 and t 2 ; by observation, for the special
case of a square wave, t1 = t b e g i n , = t 2 = t e n d , and HIC = GSI. PROBLEM FORMULATION AND ANALYSIS
For other acceleration-time profiles, however, the correlation The expression for HIC given in Eq. 2 m a y be written in the
is not obvious. A n injury threshold value of 1000 is main-
alternate form
tained for HIC, and it is to be stressed that t1 and t 2 are n o t
simply the values of time at the onset and end of the impact,
but are the values, within these bounds, which maximize HIC.
Evaluation of the GSI and HIC numbers for an experimental
acceleration-time profile is generally accomplished by numeri-
cal analyses, since an analytical expression for a(t) in Eqs. 1 Since a(t) is a real-world physical acceleration profile, it m a y
and 2 is not available. The academic problems of evaluating be assumed continuous and integrable in the time domain
the equations for simple pulses m a y be completed with little 0 < t < T where T is the total duration of the acceleration pulse.
difficulty. The limits of integration t1 and t 2 are defined to be within
The GSI number m a y be calculated for a complex pulse by
straightforward numerical integration, and the patient analyst this time domain, and therefore HIC is a function of these
can obtain reasonable results without the aid of a computer. limits. One can write
Relatively simple analog-type electrical instrumentation has
also been developed to produce strip chart recordings of GSI
versus time from onset of impact. Evaluation of the HIC num- where f is theright-handside of Eq. 3 and is differentiable.
ber, however, is not accomplished with such ease. The maxi- T o evaluate the HIC number, it is necessary to determine the
mizing process to evaluate the integration limits t1 and t 2 i n extremum values of HIC in Eq. 4 by mathematically consider-
ing a function of two variables t1 and t 2 in a region R b o u n d e d
Eq. 2 renders longhand or analog techniques wholly imprac-
tical for anything but simple pulses. by 0 < t 1 < T and 0 < t 2 < T .
Numerical evaluation of the HIC number requires analog-to- Let it be assumed that f(t1, t 2 ) possesses a m a x i m u m at
digital (A-to-D) conversion of the acceleration-time profile some interior point (t1, t 2 ).* The theory of analysis of func-
using a sampling rate sufficient to characterize the pulse ac-
tions of two independent variables then requires that the
curately. These data are easily processed by a digital com-
puter. The HIC number is determined by evaluating Eq. 2 for differentiable function f(t1, t 2 ) satisfies the conditions
all iterative combinations of the integration limits t1 and t2.
If the A-to-D sampling rate is sufficiently rapid, one need only
use the sample times for candidate t1 and t 2 values. If the
sampling rate is slow, for example, when the acceleration-time
profile has been approximated by manual (visual) selection of *This assumption is good tor all physical acceleration pro-
straight-line segments spanning various time intervals, then it files of interest. Since 0 < t < T defines the total duration of
is necessary to define a time increment At and consider all the pulse, one can write a(0) = 0 = a(T). In addition, since all
values to t = tb egin + n At (n = 0,1,2 ...) falling within the pulses of interest have finite magnitudes and durations, there
is an infinitesimal error in assuming that a(0) = 0 in an epsilon-
pulse as candidates for t1 and t2. Sampling rates or At incre-
neighborhood (e > 0, arbitrarily small) about t = 0 and t = T.
ments of the order of 1000-10000/s or 0.1-1.0 m s are typically With these conditions in force, it is easily observed from Eq. 3
used in automotive crash testing where the acceleration pulse that the m a x i m u m occurs at an interior point. The proof is as
durations generally range from 50-200 ms. follows: consider the interior point (t1, t 2 ) = (a, T - j3) where
N o analytical study, using functional analysis, to analyze the a > 0, (3 > 0, and a < T - (3. a(t1) = 0 if a < e, and a(t12)= 0 if
maximization requirement for the HIC has appeared in the (3 < e. Let j3 take on any admissible value and require a < e.
literature. The primary objective of the following presentation Then f(0, T - f) < f(a, T - 0) because the integral in Eq. 3 has
is to consider the mathematical problem of describing the be- -1.5
havior of HIC, defined by Eq. 2, through analysis of a function the same value for t1 = 0 and t1 = a, but(t2-t1)- is
of two independent variables t1 and t 2 . The analysis is carried smaller for t1 = 0 than for t1 = a. Similarly, let a take on any
out from a general point of view to establish the conditions for admissible value and require j3 < e. Then f(a, T ) < f(a, T - j3)
maximization of HIC. Illustrative examples are provided by because the integral in Eq. 3 has the same value for t 2 - T and
applying the established conditions to simple pulses, including
the half-sine, triangle, trapezoidal, and square wave. In each t 2 = T - 0, but (t2 - t1)- 1.5 is smaller for t 2 = T than t 2 =
case, the limits of integration t1 and t 2 are determined, and T - (3. Thus, if either t1 or t 2 is a boundary point, f(t1, t 2 ) is
the HIC and GSI numbers are calculated from derived expres- not a m a x i m u m .

399
 Downloaded from SAE International by University of New South Wales, Wednesday, September 19, 2018

400 C. CHOU AND G. NYQUIST

at (ti,t2). A physical interpretation of Eq. 5, where there is where the subscripts denote differentiation with respect to
a maximum, can be visualized as shown in Fig. 1. It should be those variables, then f(t1,t2) has a maximum at (t1,t2) if
pointed out, however, that the conditions of Eq. 5 are neces New Text
ft1t1 <0;f(t1,t2) has a minimum at (t1,t2) if ft1t1 > 0.
sary but not sufficient for a maximum or a minimum. One
could have a saddle point (Fig. 2) with Eq. 5 in force. The The above theorem enables one to pursue the extremum
saddle point is neither a maximum nor a minimum. value of HIC. Differentiating f(t1,t2) with respect to t1 de
Points on f(t1,t2) that satisfy the conditions of Eq. 5 are fined by the right-hand side of Eq. 3 gives
called critical points. To determine whether a maximum,
minimum, or neither exists at a critical point, it is necessary
to examine the behavior of the second derivatives in the
neighborhood of the point. The following theorem (4) is
applicable:
Let f(t1,t2) have second derivatives in the region including
the point (t1,t2). If

and
Differentiation of the integral in Eq. 7 may be accomplished
using Leibnitz's rule (4), and it follows that
 Downloaded from SAE International by University of New South Wales, Wednesday, September 19, 2018

STUDIES O F H E A D INJURY CRITERION (HIC) 401

Applying Eq. 8 to Eq. 7 and setting ∂f/∂t1 equal to zero yields

Similarly, performing partial differentiation of f(t1, t2) with

respect to t2 yields

where the H I C is maximized, a(t2) = a(t2) and the average

acceleration between t1 and t2 is 5/3 of the acceleration at t1
Eqs. 9 and 10 provide two simultaneous equations in the or t2.
two unknowns t1 and t2- Theoretically, this system can be
The usefulness of the above analyses, particularly the result
solved to obtain the critical points (t1 ,t2) of f(t1,t2). De­ expressed by Eq. 12, will be illustrated in some examples.
pending on the complexity of the acceleration function a(t),
solving Eqs. 9 and 10 m a y be a difficult task. Even if the sys­ ILLUSTRATIVE E X A M P L E S
tem cannot be solved, Eqs. 9 and 10 are still of considerable
usefulness since they enable important results to be realized. HALF-SINE PULSE - Consider the half-sine acceleration
Subtraction of Eq. 10 from Eq. 9 gives, at critical points, profile shown in Fig. 3. The acceleration pulse a(t) is de­
scribed by the expression

Since for a nontrivial solution t2 ≠ t1, Eq. 11 implies that

Eq. 12 tells us that the values t1 = t1 and t2 = t2 which maxi­

mize HIC (see Eq. 2) are such that the acceleration at t = t1 is Substituting a(t) from Eq. 13 into Eq. 2 gives
equal to the acceleration at t = t2.
Another result m a y be realized by rewriting Eqs. 9 and 10 in
the following forms:

Instead of direct application of Eqs. 9 and 10, the problem

will be treated in a somewhat different manner. Note that the
acceleration pulse is symmetrical with respect to t = T/2. This
enables Eq. 14 to be reduced to a function of one real variable
t1 after some manipulation.
From Eq. 12, it is k n o w n that a(t1) = a(t2) when t1 and t2
are selected so as to maximize HIC. For a pulse symmetric
The former equation holds when ∂f/∂t1 = 0, and the latter about t = T/2, this implies that
holds when ∂f/∂t2 = 0. The left-hand sides of the equations
are identical and define the average acceleration between t = t1
and t = t2. The equations tell us that for any value of t1
and therefore
where ∂f/∂t1 = 0, the average acceleration between t1 and an
arbitrary value of t2 is equal to 5/3 of the acceleration at t1.
Similarly, for any value of t2 where ∂f/∂t2 = 0, the average
These observations make it clear that
acceleration between t2 and an arbitrary value of t1 is equal
to 5/3 of the acceleration at t2> Thus, at (t1, t2) = (t1,t2),
 Downloaded from SAE International by University of New South Wales, Wednesday, September 19, 2018

402 C. CHOU AND G. NYQUIST

and this implies a m a x i m u m point, since w e have already required

the first derivative to vanish.
The GSI defined by Eq. 1 for the half-sine acceleration pulse
of Eq. 13 is given by the relation

Therefore, Eq. 14 m a y be written in the equivalent form

The integral in Eq. 22 cannot be evaluated in terms of elemen-

tary functions; however, it can be expressed in terms of g a m m a
functions. It can be shown that Eq. 22 m a y be reduced to
Carrying out the integration in Eq. 16, it takes the form

and using a tabulation of g a m m a functions (6) enables evalua-

To find the extremum values of HIC in Eq. 17, it is neces- tion of the right-hand side to yield
sary to determine the critical points by differentiating the
right-hand side with respect to t1 and equating the expression
to zero. This leads to the equation
Comparing Eqs. 21 and 23, it is evident that for a half-sine
pulse

Introducing a new variable

and therefore the HIC number is 9.4%smaller than the GSI.
T R I A N G U L A R P U L S E - Suppose the acceleration pulse has
a triangular profile, as indicated in Fig. 4. Mathematically, this
pulse m a y be represented by the expressions
enables Eq. 18 to be written in the form

and
which is a transcendental equation that can be solved graphic-
ally or numerically. Using a numerical method (5), the com-
puter indicates that

and therefore, from Eq. 19,

and t2 can be determined from Eq. 15 accordingly.

Substituting t1 from Eq. 20 into Eq. 17 yields The H I C equation for the triangular pulse is obtained by sub-
stituting a(t) from Eqs. 25 and 26 into Eq. 3. F r o m Eq. 12, it
is k n o w n that a(t1) = a(t2)by observation, from Fig. 4, it is
clear that this implies 0<t1 < k T and k T < t 2 < T . Thus, the
integral in Eq. 3 m a y be expressed as follows:
It can easily be shown that Eq. 20 locates a m a x i m u m point
and that the desired HIC number is therefore given by Eq. 21.
The second derivative of HIC in Eq. 17 with respect to t1 is
negative; the theory of functions of one variable tells us that
 Downloaded from SAE International by University of New South Wales, Wednesday, September 19, 2018

STUDIES OF HEAD INJURY CRITERION (HIC) 403

It can easily be shown that at the critical point of Eq. 29,

the determinant D of Eq. 6 is negative and &#x02202;2HIC/&#x02202;t1&#x02202;t1
(t1,t2)<New
0. Text
Therefore, this is the required maximum point;
from Eq. 28 the HIC number is

which may be written in the equivalent form

These integrals are easily evaluated, and one can then write

Eq. 30 indicates that the HIC number is independent of the

skewness parameter k. Only the peak acceleration Ap and
pulse duration T are involved. In addition, it is interesting to
note from Eq. 29 that

The HIC number may be established by evaluating the right

hand side of Eq. 28 using values of t1 and t2 that maximize Thus, even though the critical values t1 and t2 depend on the
the expression. It was shown earlier that the maximum will skewness parameter k, the length of the interval pertinent for
occur at a critical point t1 = t2, t2=t2 established by simul HIC number evaluation is independent of the skewness.
Using Eqs. 1,25, and 26, the GSI for the triangular pulse
taneous solution of Eqs. 9 and 12. Again, the integrals in these may be expressed as
equations may be evaluated by using Eq. 27. The results of
evaluating Eqs. 9 and 12 are

and The integrals of this expression are easily evaluated, and the
result is

A simultaneous solution of these equations gives the critical or

points. One obtains the quadratic equation

which has the roots Comparing Eqs. 30 and 31, it is evident that for any tri
angular pulse

The root t1 = kT leads to t2 = kT. This critical point is the and therefore the HIC number is 13.8% smaller than the GSI.
minimum HIC = 0 and is therefore discarded from further con TRAPEZOIDAL PULSE - A trapezoidal acceleration profile
sideration. The other root leads to (Fig. 5) may be analyzed by following the same procedure
used in the preceding triangle example. The trapezoid can be
expressed analytically by the expressions
 Downloaded from SAE International by University of New South Wales, Wednesday, September 19, 2018

404 C. CHOU AND G. NYQUIST

Applying Eq. 36 to Eq. 2 yields

and Eq. 34 is eliminated if k2 = 1. The quantity A p s the

As in the previous example of a triangular pulse, the H I C
plateau acceleration. Constants k1 and k2 characterize the
number m a y be obtained by evaluating the right-hand side of
steepness of the ramps, and T is the pulse duration. Eq. 37 using values of t1 and t2 that maximize the expression.
Our aim is to evaluate Eq. 2, which defines HIC. The inte- The m a x i m u m occurs at a critical point t1 = t1, t2 = t2 estab-
gral in the equation can be dealt with if the proper expressions
lished by a simultaneous solution of Eqs. 9 and 12. Eq. 36
for a(t) between t1 and t2 can be determined. The determina- m a y be used to evaluate the integrals in these equations, and
tion m a y be m a d e by using the condition a(t1) = a(t2)estab- Eqs. 32, 34, and 35 enable evaluation of a(t1) and a(t2). The
lished in Eq. 12 and by observing that theright-handside of results of evaluating Eqs. 9 and 12 are the expressions
Eq. 2 is simply the average acceleration between t1 and t2,
raised to the power of 2.5, multiplied by the time duration
(t2 - t1). For any values of t1 and t2 on the plateau portion
of the trapezoid, the average acceleration is constant ( A p ) , and
therefore H I C increases as t1 diminishes and t2 increases. Thus
and
if t1 and t2 are on the plateau, they must be (t1, t2) = (k1T,
k2T) to maximize HIC. In addition, since Eq. 12 requires
a(t1) = a(t2), then t1 not on the plateau implies that t2 is not
on the plateau, and vice versa. The conclusion from this dis-
cussion is that the t1 and t2 values of interest are on the A simultaneous solution of these equations gives the critical
ramps; that is, points. One obtains the quadratic equation

and which has the roots

The relations in Eq. 35 enable proper selection of a(t) from

Eqs. 32-34 for completing the integral in Eq. 2, and one can
write Substituting the values of t1 for t1 in Eq. 38 gives the cor-
 Downloaded from SAE International by University of New South Wales, Wednesday, September 19, 2018

STUDIES OF HEAD INJURY CRITERION (HIC) 405

responding values of t2:

The generality of the analysis must be restricted at this point

to proceed further. One must consider some specific selec­
tions of k1 and k2. It is interesting to note that if k 1 = k 2 = k
the trapezoidal acceleration profile of Eqs. 32-34 reduces to a
triangle. Also, Eqs. 39 and 40 yield the critical point (■=- kT,
4 + 3k \
The GSI for the trapezoid defined by Eqs. 32-34 may be
— - — T J , which coincides with the result obtained in the pre­ evaluated by setting (t1, t 2 ) = (0,T) and raising a(t) to the
vious example.
power of 2.5 in Eq. 36. This gives
The special case where k2 = 1 - k1 will be examined next.
This is an isosceles trapezoid. Eqs. 39 and 38 reduced to

and

which is easily evaluated to get

respectively. Then, from Eq. 37, the HIC number is given by
the relation

NUMERICAL RESULTS AND DISCUSSION

The analytical results derived for the various simple accelera­

where t1 is evaluated using Eq. 41 with the sign that maxi­
tion profiles are summarized in Table 1. Half-sine, trapezoidal,
mizes HIC. triangular, and square-wave pulses are included; however, no
The solution for a square-wave acceleration profile (Fig. 6) HIC number equations are presented for the trapezoid because
can be obtained as a special case of the trapezoidal solution by the form of the results depends on the values of the parameters
letting k1 — > 0 and k2 — ► 1 .* Eqs. 39 and 40 give t1 = 0 and k1 and k 2 used in defining the shape of the pulse. The mathe­
t 2 = T, respectively. The right-hand side of Eq. 43 reduces to matical expression for acceleration is shown directly below
T A p , which by Eq. 1 is also the GSI. A n analytical proof each acceleration profile sketch, and the expressions for ve­
locity change (∆V), displacement (S), and pulse duration (T)
has been presented, therefore, for the observation made in the are included for convenience. Following this information are
beginning of the paper that HIC = GSI for a square wave. the expressions for the critical time values (t1, t 2 ) maximizing
HIC, the equations for the HIC number and GSI, and finally
*For a square-wave acceleration profile a(t) = A in the do­
the percent reduction by which the HIC number is less than
main 0 < t < T , HIC = Ap- 5 (t2 - tj). Since thefirstderivatives the GSI. The HIC number and GSI equations have been writ­
of HIC with respect to t1 and t 2 do not vanish, the necessary ten in two different forms for convenience. One m a y use
requirements given in Eq. 5 are not satisfied. Noting that the either A and T or A and A V as independent variables.
HIC is a monotonic increasing function of t 2 f° r a fixed ti anc Numerical computations have been carried out using the
a monotonic decreasing function of t1 for a given t 2 , it thus equations presented in Table 1. The results are tabulated by
listing the pulse duration (T), displacement (S), HIC number,
attains a m a x i m u m at (t1, t 2 ) = (o, T). The solution for the and GSI for several prescribed combinations of velocity change
square wave may be also obtained nicely, however, by taking (∆V) and peak acceleration (Ap). Data for the half-sine accel­
the limiting case of the trapezoidal solution; that is, k1 — ► 0 eration profile are shown in Table 2; the isoceles triangle and
and k 2 —* 1. square wave are covered in Tables 3 and 4, respectively.
 Downloaded from SAE International by University of New South Wales, Wednesday, September 19, 2018

406 C. CHOU AND G. NYQUIST

The above data are represented graphically in Figs. 7-9, and change. The displacement follows the (?V)2 law; that is, it
a convenient semilogarithmic composite plot of HIC number varies directly with the square of the velocity change.
and GSI versus peak acceleration is presented in Fig. 10 for the It is remarked that five parameters are involved in the data
three pulse shapes. The curves in Fig. 10 hold for a pulse dura of Figs. 7-9. For any one of the acceleration profiles, if two of
tion of 1 ms. To determine the HIC number or GSI for a pulse the five parameters are known, the remaining three can be de
of longer duration, simply multiply the 1 ms answer by the termined by direct analysis of the figures.
duration expressed in ms. Looking
back at the developments presented in the previous
Some pertinent comments can be made upon viewing Figs. pages, it is apparent that the mathematical technique for de
7-9. From the upper diagram of each figure, it is apparent that termining the HIC number is practical if the acceleration pulse
the HIC number and GSI are linearly related to the velocity is analytically tractable. Although the method can be applied
change. Second, the pulse duration is also a linear function of to general cases, the determination of extremum values, in
velocity change. The lower diagram of each figure indicates a general, is of considerable complexity if the acceleration pulse
nonlinear relationship between displacement and velocity is not of a simple form.
 Downloaded from SAE International by University of New South Wales, Wednesday, September 19, 2018

STUDIES OF HEAD INJURY CRITERION (HIC) 407

 Downloaded from SAE International by University of New South Wales, Wednesday, September 19, 2018

408 C. CHOU AND G. NYQUIST

Downloaded from SAE International by University of New South Wales, Wednesday, September 19, 2018

409
 Downloaded from SAE International by University of New South Wales, Wednesday, September 19, 2018

410 C. CHOU AND G. NYQUIST

CONCLUSION is also due to Dr. A . I. King of Biomechanics Research Center

for his comments through reading of the original manuscript.
1. The HIC number and GSI are equal for a square-wave
acceleration pulse. For other acceleration profiles, and HIC
number is always smaller than the GSI. For a given peak ac- REFERENCES
celeration and time duration, the HIC number for a half-sine
pulse is 9.4%less than the GSI; for a triangular pulse it is
13.8% less. 1. C . W . Gadd, "Use of a Weighted-Impulse Criterion for
2 . The HIC number and GSI are independent of the skew- Estimating Injury Hazard." Paper 660793, Proceedings of
ness (k) of a triangular pulse. The displacement, however, de- Tenth Stapp Car Crash Conference, P-12. N e w York: Society
pends on k. of Automotive Engineers, Inc., 1966.
3. For any arbitrary continuous acceleration profile, regard 2 . L . M . Patrick, H . R . Lissner, and E . S. Gurdjian, "Survival
less of shape, the acceleration magnitudes are equal at each by Design—Head Protection." Proceedings of Seventh Stapp
end of the unique time interval used to evaluate the HIC n u m Car Crash Conference. Springfield, 111.: Charles C Thomas,
ber. That is, using the terminology of the mathematical pre- 1963.
sentation, a(tj) = a ^ ) . This conclusion significantly reduces 3. Department of Transportation N H T S A Docket Number
the number of trial pairs (t j , t2) necessary in evaluating the 69-7, Notice 19, Occupant Crash Protection-Head Injury Cri-
teria, S6.2 of M V S S #208.
HIC number by computer.
4 . F . B . Hildebrand, "Advanced Calculus for Applications."
4 . The average acceleration between t\ and U is 5/3 times
Englewood Cliffs, N . J.: Prentice Hall, 1962, pp. 350-351, 360.
the acceleration at tj or tj. 5. P . Henrici, "Elements of Numerical Analysis." N e w
York: John Wiley, 1964.
ACKNOWLEDGMENTS 6. M . R . Speigel, "Mathematical Handbook of Formulas and
The authors wish to thank Prof. L . M . Patrick of Biome- Tables." Schaum's Outline Series, N e w York: McGraw-Hill
chanics Research Center for suggesting the problem. Gratitude Book C o . , 1968, p. 235.