Engineering Aspects of Human Skull Fracture

-Experimental Study & Theoretical Consideration On

Brain Damage-

Hideaki MASUZAWA*, Kimiyoshi HIRAKAWA, Norio NAKAMURA and Keiji SANO

Department of neurosurgery, Faculty of medicine, University of Tokyo,

Masatomo KIHIRA and Tsuyoshi HAYASFII

Department of Aeronautics, Faculty of Engineering, University of Tokyo,

Japan

Japan

Abstract

1) Four dry human skulls were statically compressed in fronto-occipital direction. The forces to produce bone fracture ranged between 600 and 1000 kg weight. The load-deflection curves gave spring rates of 400 to 900 kg/mm, average 600 kg/mm.

2) Semi-free fall test to give impact on twenty dry human skulls at mid-frontal regions were performed using a strain-gauge load cell, an accelerometer, DC amplifiers and a cathode-ray oscilloscope. Various buffer materials including the scalp simulator were used. At fracture notching and abrupt lowering of the load curves as well as high frequency vibration of the acceleration curves were noted. The rise time at fracture became shorter compared to those when fracture did not occur. The dynamic load level for frontal bone fracture was between 400 and 1300 kg, was independent of acceleration, energy, buffer materials as well as rise times. The peak loads at the moment of fracture were closely related to the weights of dry skulls. These results indicate that human skulls have the characteristics of brittle fracture both statically and dynamically. 3) Based on the above conclusions the tolerance limits against brain damage

was theoretically considered and the Wayne State Human Tolerance Curve was criticized.

Postulating a head falling brow-down on a hard flat surface without the effect of the body, the threshold for fracture in terms of peak acceleration and rise time is very close to the Wayne State Human Tolerance Curve (H. T. C.). Above this threshold or even when higher energy is applied, the head does not give rise to a higher acceleration but fracture should occur at the same peak acceleration and

possibly shorter rise time. This means transition to the safe zone of the H. T. C. Clini-cally, however, severer brain damage is expected. The H.T.C. and the pressure gradi-ent theory due to translational acceleration should not be applied to brain dama-

ge in the presence of fracture. Instead, skull deformation and possibly snap-back of

* Present address: Department of neurosurgery, Fuchu General Hospital, Tokyo

-46-

the deformed skull should be considered as the mechanism of the brain damage associated with skull fracture.

CLARIFICATION OF THE HUMAN TOLERANCE LIMITS against trauma would promote prevention of injury as well as application in clinical medicine. In recent years several papers have appeared concerning with the mechanism of head injury. Inmost of theml>2>, however, skulls were considered merely as a non-crush-able rigid container. Clinically skull fracture has much to do with the severity of head injury. Acute epidural hematoma for example is closely related with fracture. Ommaya et a13) suggested intimate relationship present between fracture and brain contusion. Skull fractures are mostly either 1) linear fractures due to blunt impact or 2)

depressed fractures due to sharp impact. The former is more frequently encountered in everyday life. The mechanical limits of skull fracture were reported by many researchers. Most of them4>5>, however, treated energy as the limiting factor in skull fracture. In 1968 Thomas et a16) reported that human cadaver heads were fractured by frontal static loads at 1150-2200 lbs and by dynamic loads at 940-2040 lbs. Nahum et a17> showed that fracture was determined only by the peak of the load and not by its duration and that the threshold for frontal impact fracture was 630-1990 lbs. These reports clearly pointed the impact load as the threshold for fracture, but still the individual variations of thresholds remained unsolved.

The following mechanical experiments were conducted in order to find what kind of engineering parameters determine the threshold for human skull fracture; what are the backgrounds for the individual variations; and how the threshold for skull fracture are concerned with the brain damage.

STATIC LOAD TEST OF HUMAN DRY SKULLS

Four dry human skulls imported from India were used. Skulls were com-

pressed statically in fronto-occipital direction as seen in Fig. 1. The loading head was made of brass, 4 by 4.5 cm in size, nearly oval in contour and concave in surface. This was glued with methyl metacrylate on the midsagittal frontal area of skulls 5.5 cm apart from the nasion. Two skulls were compressed gradually, and the loads and deflections were read until they were fractured. Three dial-gauges were used for the deflection. Two other skulls were loaded quasi-statically at the rate of 400 kg/min using a KYOWA LC-5TA strain-gauge load cell and two electric dial-

gauges. These outputs were fed into KYOWA DM-6N strain-amplifier and then recorded on KYOWA RMV-33H Rapicorder paper oscillograph. Fig. 2 shows the load-deflection curves of the four skulls. Fractures were multi-

linear centering at the frontal loading area and often extending to the skull base

(Fig. 3). DISCUSSION: The loading head has about 14 square cm of area, and is for beyond the threshold of 6.5 square cm between depressed and linear fractures

-47-

in adult human skulls, as stated by Holcombe and Herod8).

The loads at fracture ranged between 600 and 997 kg. and well corresponded

to the results of the static load test by Thomas et al6). This would imply that the

static load limits of skulls are not much different whether dry or wet, and whether

-48-

Fig. 1

C, D

Static

Load

compression test of human skulls.

ing heads of the machine; GI, G2

S: Skull; A,

Dial gauges

B: Loading heads;

Fig . 2 Load-deflection curves of the statically compressed human skulls.

Fig. 3 An example of the statiscally fractured skull.

covered by dura mater as cadaver heads or not . Katayama9) found no statistical differences in Young's modulus and bending strength between fresh and dry bones.

Deflections just prior to fracture were 1.1-1.6 mm and about 0.6-0 .9% of the longitudinal diameter of the skulls. After destruction the fracture lines diastased and the loads diminished and macroscopic deformations occurred.

The load-deflection curves depicted in Fig. 2 were linear before fracture. This indicates that skulls have the characteristics of brittle elasticity at the antero-

posterior compression. The spring constant for each curve was estimated as 420, 490, 610 and 860 kg/mm, average 600 kg/mm . SUMMARY: i) Compression of four dry human skulls antero-posteriorly

resulted in multilinear fractures extending to the base . ii) The loads at fracture ranged between 600 and 1000 kg, the deflection just prior to fracture 1.1-1.6 mm. After fracture, skulls showed diminished loads and macroscopic deformations. iii) The load-deflection curves were linear , indicating that skulls are brittle and the

spring constants were calculated as 400-900 kg/mm, average 600 kg/mm.

DYNAMIC LOAD TEST OF HUMAN SKULLS

PROCEDURE: Fig. 4 illustrates the semi-free fall method to give the dy-namic load to skulls . Twenty dry human skulls were used. Skulls were hanged so as to receive impact at the mid -frontal region, 4.5-5.5 cm apart from the nasion.

-49-

The hanging wire was passed through a pulley at the ceiling and the other end was fixed. The skull was raised to the desired height and was released instantaneously with a magnetic switch. Skulls No. (1)-(7) were fitted with KYOWA AS-1000A uniaxial strain accelerometer which was glued on the occipital region of skulls. The dynamic loads were measured by a KYOWA LC-5TA strain-gauge load cell with a metallic saucer attached on the top. Various shock-absorbing or buffer materials were placed on this saucer to receive falling skulls. Fracture of skulls were exami-ned after every fall and graded as: (-) no fracture, (1+) hairline fracture, (2+) single or multiple linear fractures without permanent deformations, and (3+) mul-tilinear or comminuted fractures with grossly displaced bone fragments. Skulls No.

(1) to (7) were dropped by gradual increments of heights. Skulls No. (13) to (26) were dropped mostly twice, first below and secondly above the pre-estimated thre-shold for fracture. The outputs from the load cell and accelerometer were fed into NIHON KOHDEN AVH-2 high-gain DC amplifier, and the resulting signals were

played on NIHON KOHDEN VC-7A cathode-ray oscilloscope and recorded on polaroid films as in Fig. 9.

HUMAN SKULLS: Twenty adult skulls were used. Bone weight, a-p and la-teral diameters were measured, rubber bags put into the cavity, 5% aquaous gelatin solution with formalin was filled into the bag under pressure, the foramen magnum

-50-

Fig. 4 Dynamic load test of human skulls.

Fig. 5 An example of the dynamic load test .

Fig. 6 Block-diagramm of the dynamic load test.

was sealed with rubber and the whole or total weight was again measured . Some skulls were filled with the mixture of gelatin and red lead to increase the whole weight. As seen in Fig . 5 some skulls were also equipped with an aluminum plate to function with the photo-transistor . BUFFERS: Table I lists the various buffers placed on the metallic saucer

of the load cell. Among these were included SEKISUI Chemicals foamed stylene of 17 mm in thickness as the human scalp simulator in buffer capacity as proposed by one of the authors (H . M.)10).

- 51 -

Fig. 7 Vibratory responce of the load cell system.

Table 1 Description of Skulls and Buffers

-52-

Fig. 8 Skull No. 3 after the dynamic load test.

MEASUREMENTS: Fig. 6 shows the block-diagram of the instrumentation system. Both the load cell and the accelerometer were strain-gauge type. DC ampli-fiers were used to obtain hi-fidelity over the high frequency range. The DC source was hand-made with dry batteries , variable resistances and voltage meters. Calibra-tion on the oscilloscope was deduced by calculation . A photo-transistor was used as the external trigger for the sweep. Load cells are made for use in static tests. Fig. 7 shows the impact responce of the load cell fixed to the base and with a saucer on it . Some 3000 Hz wave is seen mixed with higher frequencies . This would indicate that the load cell system are capa-ble of measuring waves of up to 1000 Hz .

A HITACHI high speed camera Model 16HD set at 8000 fps was driven at the falls of skulls No. (16) to (20). The films were analyzed by means of a NACK motion analyzer Model 16S to obtain the falling speed just prior to hit. These were 90%, 94%, 99% and 94% respectively, of the theoretical speeds considered as free falls. RESULTS: Fig. 8 demonstrates an example of a fractured skull and a frac-

- 53 -

Fig. 9 Oscillograph records of the dynamic load test.

ture line extending to the orbit. In Fig. 9 the load-time and acceleration-time curves of the same skull are seen. The load curve at fracture appears less smooth and notched probably at the moment of fracture. The acceleration curve shows high frequency oscillation coincident with fracture. Fig. 10 shows the same oscillations of the acceleration as well as abrupt lowering of the load curves. Shown in Table 1 is the description of skulls and buffers used. In Tables 2-1 and 2-2 are listed the results of the dropping tests including peak values of the acceleration (Peak Aocc) and/or the load (Peak Load), rise times of the load curves and the grade of frac-ture. Energy was calculated as the drop height (Drop ht.) multiplied by the total weight of the skull. DISCUSSION: Table 2-1 shows variable values of the accelerations at fracture and indicates that fracture is not determined by the acceleration value itself. Fig. 11 shows skull fractures by frontal impact occurring at peak loads of

400-1300 kg. This is in accordance with the results of the static test. This is also in accordance with that of Nahum et a17) and Thomas et alb) both using cadaver

-54-

Fig. 10 Oscillograph records of the dynamic load test.

heads. It would mean that the dryness and the absence of dura mater and other tissues around skulls hardly change the dynamic capcity of skulls.

Correlations between peak loads at fracture and bone weights and the ratios of a-p versus lateral diameters of skulls were seen in Fig. 12. The correlation of the latter is not so significant whereas that of the former is statistically significant (at the risk of 1%). Therefore the individual variations as formerly thought should be mainly due to the difierences of bone weights. This holds true in various buffer conditions. The shaded circles in Fig. 12 indicate the results of the falling on the human scalp simulator and the other marks those of other buffer conditions . Furthermore, it is clear that this experiment is not concerned with minimal falling energy to cause fracture. Note those tests of skulls No. (13) to (26) where above-threshold high-energy falling distances were experimented . Even under these conditions the peak loads at fracture seem to be predetermined by their bone weights. In other words, every skull has its own fracture peak load according to its own bone weight. This is not influenced by buffer conditions or by length of the rise time. It can be said Theologically that skulls should have the characteristics of brittle fracture .

-55-

The falling energies at fracture are variable depending on the buffer conditions as seen in Fig. 11. When, however, falling on the foamed stylene of 17 mm in thi-ckness or the human scalp simulator, most of skulls were fractured with the energy of over 3-4 kg- m. The weight of a human head itself is about 4-5 kg. Therefore if a head is falling by itself, it should be fractured at a fall of below one meter of height. The skulls No. (3) and (5) weighed 4.4 kg and 4.8 kg respectively and were fractured at the drop heights of 1.5 and one meter respectively on the scalp simu-lator. These data suggest that the threshold for fracture of a human head falling by itself on a hard flat surface should be around one meter fall in height. This seems not so far from the condition in which a man topples abruptly on pavement. Fig. 13 shows tendency of rise times to become shorter when higher impact

loads are generated or when farctures occurred. We adopt the value of the rise time at falling distance of one meter on the scalp simulator without fracture as the stan-dard rise time of the particular skull and under the particular buffer condition.

- 56 -

Table 2-1 Results of Frontal Skull Impacts

Fig. 14, then, shows the relationship between the fracturing loads and the standard rise times. It is clear from this figure that the heavier the skull bone, the shorter the standard rise time is.

Skull No. (7) wore a helmet when dropped and exhibited very high fracturing load value. This appears to be inaccordance with the fore-mentioned statement that the fracturing load is pre -determined . This is explained as follows: the hard outer shell of a helmet distributes the load and lessens the stress on unit areas . This load-distributing effect should raise the load limit for fracture . The load-time curves of the skulls in Fig 9 and Fig. 10 are noted to fall ab-

ruptly at the moment of fracture . Rise times become shortened or at least remain unchanged as noted in Tables 2-1 and 2-2. This implies that the products of loads and times become smaller with fracture . The products of accelerations and times should also become smaller. If one takes it true that brain damages are effected by the multiplication of translational acceleration and time duration 11) 12)13) one could

- 57 -

Table 2-2 Results of Frontal Skull Impacts

expect decrease of brain damages by skull fracture. SUMMARY: i) Semi-free fall test of twenty human skulls were conducted

to give frontal impact. The fracturing limits were determined by impact loads and not by impact accelerations or falling energies. ii) The fracturing loads ranged be-tween 400 and 1300 kg. This seemed correlated with dry bone weights indepen-dent of buffer conditions. The heavier the skulls are, the higher the fracturing loads

-58-

Fig. 11 Peak load-energy relationship

indicates foamed stylene of 17 mm

other materials.

of the dynamic load test.

in thickness. The other

The

marks

shaded circle

indicate the

Fig. 12

marks

Peak load values at

are the same as infracture in

Fig. 11.

relation with skull characteristics. The

become. iii) Skulls should be fractured when the loads reach between 400 and 1300 kg. Higher energies may cause fractures earlier and/or severer but the loads should remain the same . iv) Under the same buffer condition, the heavier the skulls are, the higher the loads at fracture and the shorter the rise times are or the more brittle they are. v) Helmets have not only buffer effects but also load-distributing effects, vi) The load curves at fracture were observed to fall abruptly and the rise times to become shorter or remain the same. This implies that fractures decrease the multiplied effect of acceleration and its duration . Therefore, it might be said that fractures should alleviate brain damages due to translational accelerations .

- 59 -

Fig. 13 Peak load versus rise time in

indicates foamed stylene of 17 mm

materials.

the dynamic load test. The shaded

in thickness and the open circle thecircle other

Fig. 14 Peak load at fracture versus standard rise time. See the text for details.

THEORETICAL CONSIDERATION ON BRAIN DAMAGE FROM THE VIEWPOINT OF THE THRESHOLD FOR SKULL FRACTURE

THRESHOLD FOR SKULL FRACTURE AND ACCELERATION: The threshold for skull fracture of a man is determined by impact load as noted above. This threshold in terms of load (F*) should be the maximum load the head could

generate under frontal impact. If fracture occurs, the rise time of the load curve becomes more or less shorter. The impact load curves under a uniform buffer condition are theoretically categorized into three types as in Fig. 15. Type A belongs to a low-energy fall without fracture and is half-sine wave in shape as seen in Fig. 9. TA is the rise time

on this condition. Type B belongs to the condition when the energy becomes higher and the load peak reaches F* just enough to cause fracture. The rise time here is the same or slightly shorter than TA. The latter half of the load curve falls ab-ruptly due to fracture. Type C belongs to a higher energy fall than type B. The load curve reaches F* before it rises to the peak. Fracture occurs earlier, the rise time becomes shorter than TA and then the load curve falls abruptly down. Fig. 14 may be considered as representing the relations between F* and TA

when one meter fall on the human scalp simulator is postulated as type A condition. The ordinate of this figure can be converted to load factor or acceleration as in Fig. 16 when the mass of a head is considered to be 5 kg. and the effects of bodies or extremities are arbitrarily neglected. Fig. 16 resembles very well the Wayne State Human Tolerance Curve (H. T. C.)111 (Fig. 17) or Human Tolerance Limit

(H. T. L.) proposed by one of the authors (T. H.)12). HUMAN TOLERANCE CURVE AND FRACTURE: Assuming that the Wayne State Tolerance Curve is similar to the skull fracture limits against a hard flat surface, then will come the conclusion that the Wayne State Tolerance Curve

- 60 -

Fig. 15

text.

Theoretical impact load-time curves. A, B, C correspond to those in the

is not valid in the presence of skull fracture as illustrated in the following theoreti-cal considerations.

1) When a head is falling on a hard flat surface without any effect of body, the head has a certain skull fracture limit in load (F*) and in acceleration (G*) . Point P in Fig. 18 is given as P (G*, TA). Skull fracture belonging to type B can be shown as point Q lying close to point P. Clinically one may expect in this case skull fracture without serious brain damage . Point Q lies indeed within the safe zone of the H. T. C.. Fracture belonging to type C can be depicted as point R which lies further on the left-hand side of point Q. This should mean transition to safer

- 61 -

Fig. 16 The

dropped on

acceleration-time relationship for

a hard flat surface.frontal skull fracture limit when

Fig. 17 The Wayne State Tolerance Curve, cited from Patrick et al").

zone of H. T. C.. Clinically, however, one may then expect severer brain damage. So there is clearly a certain discrepancy between the H. T. C. and the occurrence of brain damage in case of skull fracture.

2) When a man is falling upside-down, high values of impact load can be

generated with low acceleration because of concommittant mass effect of body through the neck. The tolerance load F* is fixed. The tolerance limit of load factor or acceleration (G') is lower than G*. The rise time may be elongated as seen in the report of Lissner et a114). Fracture belonging to type B is shown as point S in Fig. 18. Fracture belonging to type C moves to point T with shortening of rise time under increased falling energy. This point T belongs to the safe zone but can be expected to generate severe brain damage clinically.

- 62 -

Fig. 18 The H. T. C. and skull fracture. See the text for details.

Fig. 19 The H. T. C. and skull fracture. See the text for details.

3) When falling on a soft surface without the effect of body, and falling under the same height as in point Q in Fig. 18, the increased buffering of the surface will diminish the acceleration and elongate the rise time. This is shown as point U in Fig. 19 where fracture is not expected to occur due to lowered impact load. With the increased height of fall the acceleration will increase and one gets point V. If the

point V goes beyond the H. T. GG but still under the value of G*, brain damage can occur due to acceleration. Fracture is not to be expected because the load is still under the value of F*. Clinically such a case may be seen not infrequently. If the

point V goes further up to become point X, then fracture occurs as type B. Fur-ther increase in falling energy will not raise the acceleration but shortens the rise time, and the point will move to point Y. This is moving closer to the safe zone, but clinically severer brain damage is again expected. When the buffer surface is soft, a falling head sinks in and probably receives the load-distributing effects as of helmets. This will increase the tolerance limit load

(F*) as well as the acceleration (G*). One should expect some increase of the value of G* when buffering is great or impact durations long enough. Fig. 20 is drawn to show this tendency. Actually one may see the conditions 2) and 3) combined as falling on a soft sur-face upside-down. 4) When other materials are falling on or flying to a head, and the head is moving freely, linear fracture will he made by the impact load of F* and the impact acceleration of G* unless the hitting surface of the material is convex. Actually the acceleration must be smaller because the head is more or less fixed to body. DISCUSSION: From the above theroetical considerations it should be clear first that the Wayne State Human Tolerance Curve is not valid when fractures occur. Although the acceleration effect on the head is diminished with fracture, severer brain damage is clinically expected. This is suggesting that fracture itself is causing brain damage .

In monkey experiments Ommaya et a13) found frequent coup-injuries with

- 63 -

Fig. 20 The proposed revision of th e H. T. C. See the text for details.

skull fracture and proposed a theory of brain damage due to skull fracture. They attributed it to the deformation of skull with fracture. The deflection limits of skulls under static compression in fronto-occipital direc-tion were found small before fracture but after fracture macroscopic deformations were observed. In dynamic impacts enormous degrees of deflections and deforma-tions seem likely to occur with fracture. There are, however, examples to deny the direct effect of skull deformation at fracture. In children depressed fractures and comminuted fractures are known to happen easily sometimes lacerating the brain cortex by displaced bone fragments. They do have large degrees of deformations but appear to leave the brains less contused than in adults. Pediatric skulls are easily deformed but barely rebounding and linear fractures are mostly left open or diastased. On the other hand, adult skull linear fractures often rebound back to their original positions. We recently saw a case where the dura mater was tightly

pinched in the fracture line of a skull vault. These clinical facts suggests the possi-bility of brain damage caused by shell-rebounding of a skull. It is interesting to note that Gross15) speculated the cavitation mechanism by snap-back of the de-formed skull. Secondly it may be reasonable to state that there cannot exist any higher G* on human heads with the impact of fronto-occipital direction. This would also be true in case of impact in lateral or occipito-frontal directions. The highest G* value among these G*'s in all directions should be the highest possible G acceleration in human heads and this would not be far from 300 G's even in heavy bones. SUMMARY: The tolerance limit of skull fracture when a head is falling on a hard flat surface without the effect of the body is close to the Wayne State Human Tolerance Curve. This H. T. C. is valid only to the blunt acceleration injuries of the brain without skull fracture. The H. T. C. plane should be further restricted by the upper limit of acceleration in slanting slope. This upper limit of acceleration would not be far from 300 G's in heavy bones when the durations are reasonably short.

CONCLUSIONS

1) Human skulls have the characteristics of rather brittle fracture with cer-tain threshold loads. The dynamic load curve lowers rapidly at the moment of fracture. 2) With the threshold energy for fracture skull fracture may attenuate or even

save the brain damage. Energies greater than this would bring out more or less severe brain damage due to the fracture itself.

3) The limit of the skull fracture when a head is falling on a hard flat surface closely resembles the Wayne State Human Tolerance Curve. 4) Brain damages due to fracture is not explicable by the pressure gradient

theory due to translational acceleration. They may be attributed to shell rebounding or snap-back of the deformed skull on fracture. 5) The Wayne State Human Tolerance Curve is valid only to those accelera-

tion injuries of the brain in blunt head trauma without skull fracture. It is further

-64-

restricted by man heads.

the upper limit of the acceleration which could be generated on hu-

ACKNOWLEDGEMENTS

This work was supported in part

developmental scientific research No.

Automobile Research Institute grants.

by Ministry of Education granting aid for 70067 and 560067 and in part by Japan

References

1) Hayashi T.: Study of intracranial pressure caused by head impact. J. Faculty of en- gineering, Univ. of Tokyo (B), 30, 59-72, 1969.

2) Unterharnscheidt F. and Sellier K.: Mechanics and pathomorphology of closed brain injuries, Head injury conference proceedings, ed. W. F. Caveness and A. E. Walker,

Philadelphia, J. B. Lippincott, Co., 1966, pp. 321-341. 3) Ommaya A. K., Grubb R. L., Jr. and Naumann, R. A.: Coup and contre-coup injury:

Observations on the mechanics of visible brain injuries in the rhesus monkey. J. Neuro- surg., 35, 503-516, 1971.

4) Evans E. G., Lissner H. R. and Lebow M.: The relation of energy, velocity and ac- celeration to skull deformation and fracture. Surg. Gynec. Obst., 107, 593-601, 1958. 5) Kobayashi S.: Head injury and helmet. Brain & Nerve 15, 63-75, 1963 (Jap). 6) Thomas L. M., Sezgin Y., Hodgson, V. R., Cheng L. K. and Gurdjian E. S.: Static deformation and volume changes in the human skull, Proceedings of twelfth Stapp car crash conference, New York, Society of automotive engineers, Inc., 1968.

7) Nahum A. M., Gatts J. D., Gadd C. W. and Danforth J.: Impact tolerance of the skull and face, Proceedings of twelfth Stapp car crash conference, New York, Society of auto-

motive engineers, Inc., 1968. 8) Holcombe H. G. and Herod D. M.: "TRAMASAF" The development of a laboratory

instrument for trauma indication, Proceedings of G. M. safety seminar, 1968. 9) Katayama T.: Study on head injury (Experiments using human skull simulators), J. Jap. Surg. Soc. 66, 324-334, 1965 (Jap).

10) Masuzawa H.: Engineering aspects of human skull fracture. Brain and Nerve 24, 547- 561, 1972 (Jap).

11) Patrick L. M., Lissner H. R. and Gurdjian E. S.: Survival by design-head protection, Proceedings of seventh Stapp car crash conference. New York, Society of automotive engineers, Inc., 1963.

12) Hayashi T.: A theory of tolerance limit of acceleration and duration with head impact- especially the theory of Human Tolerance Curve, presented at the subcommittee meet- ing for impact injury, Japan automobile research institute, Nov. 25 1969, Tokyo.

13) Gadd C. W.: Use of a weighted-impulse criterion for estimating injury hazard, Proceed- ings of tenth Stapp car crash conference. New York, Society of automotive engineers,

Inc., 1966. 14) Lissner H. R., Lebow M. and Evans F. G.: Experimental studies on the relation between acceleration and intracranial pressure changes in man. Surg. Gynec. Obst., 106, 329-

338, 1960. 15) Gross A. G.: A new theory on the dynamics of brain concussion and brain injury. J. Neurosurg., 15, 548-561, 1958.

-65-