Anim. Behav .

, 1977, 25, 622-634

ATTACK AND DEFENSIVE BEHAVIOUR IN THE ALBINO RAT

BY ROBERT J . BLANCHARD, * D . CAROLINE BLANCHARD,

TOSHIAKI TAKAHASHI & MICHAEL J . KELLEYt
Department of Psychology, University of Hawaii at Manoa, Honolulu, Hawaii 96822

Abstract . Attack of dominant colony males of an albino rat (Rattus norvegicus) strain, on introduced
strangers, produced a non-random distribution of bites, with ventral trunk virtually never bitten .
Also, vibrissae-contact of attacker and defender interfered with bites to the defender's head and upper
back . The specific agonistic reactions of attacking and defending rats appeared to involve strategies
based on these limitations on attack : defenders utilized `boxing' and lying `on-the-back' behaviour,
which interposed ventral trunk and vibrissae between attacker and defender . In turn, the `lateral
display' permitted attackers to circumvent the defender's behaviour . Limitations on attack therefore
appeared to underlie the specific agonistic behaviour of both attacking and defending rats .

Dominant males of established wild rat (Rattus Methods

norvegicus and R . rattus) colonies have been Subjects
reported to display a host of threat and attack The six albino rat colonies used in the present
behaviours to strange male rats intruding into studies were those described in Blanchard et al .
their territories (Barnett 1963 ; Telle 1966 ; (1975) . Briefly, the colonies consisted of three
Ewer 1971 ; Alberts & Galef 1973) . Likewise, adult males and four adult females each, and
dominant male albino rats (R . norvegicus) had been formed 105 days prior to the beginning
from established colonies attacked and some- of these experiments . The colony animals were
times killed introduced strangers, although they 195 to 213 days of age during experiments 1 to 3,
showed very little agonistic behaviour toward and 415 days of age during experiment 4 . The
reintroduced members of the same colony dominant (alpha) male of each colony had been
(Blanchard et al . 1975) . The distribution of selected on the basis of magnitude of attack
lesions resulting from bites on introduced reactions in previous tests (Blanchard et al .
strangers (Blanchard et al . 1975) suggested that 1975) with introduced strangers . The introduced
some body areas, notably the back, are favoured strangers used were 61 males, with ages equiva-
attack sites . Other areas, including the entire lent to those of the colony rats at the time of
ventral surface of the rat, suffered many fewer testing. These stimulus animals, like the colony
lesions . Observations of the behaviour of the members, were Wistar derived animals from
attacking rats also suggested that attack be- Simonsen Laboratories . The intruders had been
haviour is directed : when bites were made at an singly housed for at least 30 days prior to the
animal lying with ventral surface exposed, the experiment and averaged 410 to 458 g in weight,
attacking animal usually reached under to bite for the different experiments .
at dorsal areas, or turned over its target rat to
expose the dorsal surface before biting . Apparatus
If such a non-random distribution of attack The colony enclosures were those described
sites is confirmed, it might be expected to exert a in Blanchard et al . (1975) . Briefly, these were
profound influence on the behaviour of rats in a 120 x 120 x 60-cm compartments, with ply-
conspecific agonistic encounter . Accordingly, wood walls and flooring of vinyl asbestos tile
the present studies attempted, first, to document covered with sawdust. The stimulus animals used
the existence of such a limitation on conspecific in experiments 1 and 2 were attached to a 25 x
attack in the albino rat, and to determine if this 30-cm Plexiglas platform, with screws mounted
limitation depends on differential accessibility in the Plexiglas to form a 15 x 20-cm inner
to different body loci, or the behaviour of the rectangle . The stimulus rats could be fastened
defender . Finally, these studies involved analysis to these screws by rubber bands encircling
of specific attack and defensive behaviour the fore- and hindlimbs, which prevented the
patterns of the rat, in terms of findings with stimulus animal from being turned over by an
respect to limitations on attack . attacking rat . A clear Plexiglas box, 14 X 8 x
6-cm, could also be attached to the platform .
*Reprints from R . J. Blanchard. This box was used in conjunction with a clear
tPresent address : Sussex University, Sussex, England.

622
 BLANCHARD ET AL . : RAT ATTACK AND DEFENCE 62 3

Plexiglas clamp which prevented the stimulus back, posterior back, dorsal posterior limbs,
animal from being pulled out of the box. dorsal anogenital area ; ventral loci : ventral
A Braun Nizo camera and Kodak Ektachrome aspect of head, ventral anterior limbs, thorax,
Super 8 Film were used to record the agonistic abdomen, posterior limbs, anogenital area .
sequences in experiment 4 . Two, 100-W bulbs The division between dorsal and ventral areas
in reflectors produced supplemental lighting . was made by drawing an anterior-posterior
line on each side of the animal, between the
Procedure posterior extension of the shoulder socket,
On testing days the two subordinate males of and the anterior projection of the thigh . This
each colony were removed, leaving the alpha line defined the trunk regions as dorsal or ventral,
male as the only adult male rat in each colony . while the lateral surface of the limbs were con-
Strange rats introduced into the colonies were sidered to be dorsal and the medial surface was
left in for 10-min sessions . The conditions counted as ventral . The dorsal area corresponds
under which the stimulus animals were placed in well to the surface exposed in the dorsal con-
the colonies, and the response measures taken, dition and the ventral area to the surface ex-
were different for the four experiments. posed in the ventral condition .
Experiment 1 . Attacks on dorsal or ventrally-
presented anaesthetized strangers : Wound loci. Experiment 2. Attacks on posterior-ventrally
exposed stimulus rats. In order to provide further
The stimulus rats were anaesthetized with a
40 mg/kg injection of sodium Nembutal, and data on the incidence and distribution of biting
then tied to the Plexiglas rectangle in such a attack on posterior-ventrally presented stimulus
way as to expose different portions of the body rats, experiment 2 was run using only two of the
dominant males used in the first experiment .
for the different experimental conditions . For These males were presented with a series of
the dorsal condition, the anaesthetized rat was anaesthetized strangers with only the posterior-
stretched out on the platform, lying on its ventral ventral surface exposed . The upper body of
surface, with fore- and hindlimbs pulled away these stimulus rats was concealed by an opaque
from the body and tied to the screws and its
back exposed. Rats in the twisted condition box, in contrast to the transparent box used in
had the forelimbs attached to the screws exactly experiment 1 . The two colony males used were
like the rats in the dorsal condition, but the selected on the basis of having given the highest
lower limbs were twisted to expose the anogenital levels of attack in the first experiment .
area. Rats in the ventral condition were stretched The dependent variables measured were
out on the platform on their dorsal surface with identical to those of experiment 1, except that a
the thorax and abdomen exposed . For the finer division of lesion loci was made . New
posterior-ventral condition, rats were tied down categories of lesion sites included the following :
as were those in the ventral condition, but the scrotal sac ; genital region (the penis sheath,
anterior half of the rats' bodies were covered by and the fatty tissue at its base) ; feet (un-
the Plexiglas box, exposing only abdomen and furred region of the leg and feet) ; legs (the furred
genitals and posterior limbs . portion of the animal's posterior limbs) .
The anaesthetized and tied stimulus rats were Experiment 3. Alterations of attack and de-
placed in the centre of the colony, and the fensive reactions after anaesthetization of intruder
following behaviours of the alpha male of that vibrissae . Some behaviour available to un-
colony were then observed : (1) latency to pilo- anaesthetized defenders facilitates protection of
erection ; (2) latency to bite ; (3) number of anal the head against biting attack (Blanchard et al .
sniffs (subject's nose is within 1 cm of the anal 1975) . Experiment 3 therefore described both
region of the introduced animal) ; (4) number of defender and attacker behaviour after an-
bites at the platform or box ; (5) total contact aesthetization of the defenders' vibrissae, a
time (time when the observed animal is within manipulation which appeared likely to make the
1 cm of the platform) . head more vulnerable to biting attack.
At the conclusion of the test session, the Two groups of introduced strangers were used.
stimulus animal was removed and killed with The anaesthetized vibrissae animals had had
an overdose of sodium Nembutal . The number of 0 . 3 ml of Xylocaine (2% Lidocaine (HCl (20
skin or tissue lesions on each animal were mg/ml) injected bilaterally into the vibrissae
counted and recorded by area, as follows : pad, with a 0 . 15 ml injection at the anterior
dorsal loci : head, dorsal anterior limbs, anterior medial portion of the pad, and a 0 . 15 ml
624 ANIMAL BEHAVIOUR, 25, 3

injection at the central posterior region . These movements' was added, as was a `check' category .
animals were allowed 15 min for the anaesthetic The intention movements were abrupt move-
to take effect and were then placed into the ments (at least 1 cm in magnitude) of an attack-
colony situation. In each case, vibrissae move- ing animal which tended to result in the attacker's
ments of the anaesthetized rats had ceased snout being brought into closer contact with the
before the end of the 15-min waiting period . back or sides of the other rat . If such proximity
Control animals received exactly the same was attained, a frequent result was a bite of, or at,
volume and placement of injections, but they the strange rat . The check category consisted
were injected with a normal saline solution rather of movements of the defending rat which
than the anaesthetic. tended to result in the removal of this rat's back
One rat with vibrissae anaesthetized and one or sides from the vicinity of the attacker's snout .
control rat were introduced individually into Typically, these were alterations of body orien-
each of the six colonies . During the test sessions, tation or posture, to `track' a circling attacker,
the following behaviour patterns were monitored keeping the defender's ventral surface or
and recorded for the dominant male and for the vibrissae between attacker and defender .
introduced rat : `contact time' between alpha Final categories of behaviour consisted of a
male and introduced stranger ; `flight,' observed finer division of the on-top behaviour . This
animal runs away from another animal with behaviour of standing on-top of a prone oppo-
which it has been interacting ; `chasing', observed nent was further divided into categories in
animal runs after another, fleeing rat ; `roll, which the attacker was oriented parallel to its
tumble, and bite', a vigorous movement with opponent ('on-top-parallel') as opposed to a
contact between animals and/or biting ; `boxing', perpendicular orientation (`on-top-lateral') .
a stereotyped, upright nose-to-nose posture ;
`on-the-back', the rat leans or lies on its back ; Results
`freezing', immobility with all limbs touching Attacks on Dorsal or Ventrally-Presented An-
the ground ; `rear immobile', animal assumes aesthetized Strangers : Wound Loci
an upright immobile stance, as is characteristic Table I presents each of the behavioural
of the boxing response but with no other rat measures for the dominant colony males con-
present ; `lateral display', the observed animal fronted with the four conditions of anaesthetized
orients toward another rat with its head lowered stimulus rats. These stimulus rats had dorsal or
and straight but with back and hind legs arched ; ventral surfaces exposed, or were twisted to
and the `on-top-of' behaviour in which the expose dorsal-anterior, and ventral-posterior
animal walks or stands on top of its opponent,
which is usually on its back. Each behaviour Table I . Attacks on Anaesthetized Strangers
pattern was suggested by descriptions of feral
rats (Barnett 1963), as well as by our previous Stimulus groups
observations of ongoing behaviour in the colony
situation . Ventral
Mean scores Dorsal Twisted Ventral posterior
Experiment 4. Interaction of attacker and
defender behaviour . Film sequences (18 f/sec) were Contact time (s) 579 . 33 568 .00 553 . 33 589 . 50
made for the initial 185 s period after a strange Latency to
rat was first attacked by the alpha male of each piloerection (s) 27 . 60 19 . 20 28 . 80 13 .20
colony . The films were examined in slow motion
projection and also on a frame-by-frame basis Latency to first 2 .27 2 . 15 4 . 50 6 . 60
on a film editor . The behaviours of both attacker bite (min)
and defender were classified by an experienced Anal sniffs 1 . 83 5 . 67 5 . 50 14 .00
rater, and their durations (or other appropriate
measurements) were obtained . Approximately Platform bites 16 . 00 17 . 17 27 . 50 23 .67
one-third of this data was independently scored Bites 17 . 83 13 . 83 5 .17 8 . 17
by another trained rater, with a high degree of
concordance (over 95 %) for judgments . Total bites (at 33 . 83 31 . 00 32 . 67 32 . 84
rat on platform)
The specific behaviour categories used were
those of the first experiments of this series, Lesions 12 . 33 8 . 33 3 . 50 5 . 67
with the exception that a category of `intention
 BLANCHARD ET AL . : RAT ATTACK AND DEFENCE 625

surfaces, or were covered with a box, such that did display significantly different numbers of
only the ventral-posterior surface was exposed . lesions (F = 3 df = 3 and 15, P < 0.05) .
The contact time measure makes it clear that Subsequent analyses indicated that the dorsal
all of the stimulus rats elicited approach and group had significantly more lesions than either
exploratory behaviour in the dominant colony the ventral or the ventral-posterior groups
males : during more than 90% of the 10-min (P < 0 .05 in either case). No other differences
test period for all stimulus groups, the colony in the number of lesions were significant . Thus,
males remained within 1 cm of the platform the number of lesions data is in good agreement
containing the stimulus rat. The differences in with the latency to bite data, in suggesting a
contact times for the different stimuli were not reduced tendency to bite at and wound the
statistically significant . ventral and ventral-posterior stimulus groups .
Piloerection in the colony alpha males was Locus of bites and lesions . The data on the
also a clear response to the introduced rats . number and distribution of wounds on the
Full piloerection occurred to each stimulus stimulus rats are presented in Table II . As these
condition, in less than 30 s . Differences in pilo- data indicate, the procedure of tying down the
erection latencies to the four stimulus groups stimulus animals to a platform did not com-
were not statistically significant . However, pletely preclude wounds on the concealed
analysis of variance indicated differences in surface . Our observations suggested that, especi-
olfactory investigations of the four stimulus ally for the ventral and posterior-ventral groups,
groups (F = 17 . 01 ; df = 3 and 15, P < 0.001) . the attacking rats devoted a great deal of time
Subsequent t-tests showed that more anal sniffs and effort in attempting to turn over the stimulus
were made to the ventral-posterior group, than rat, or to grasp and bite the concealed (dorsal
to any of the other groups (P < 0 .05 in each anterior) surfaces.
case) . Also, both ventral animals and twisted These data make it clear that the head was a
animals tended to receive more anal sniffs favoured attack site . Head wounds accounted
than did dorsal animals (P < 0 .05 in each case) . for more than 75 % of the total number of
The ventral group did not differ reliably from wounds in all target animals in which the head
the twisted group. was offered. These data offer a striking contrast
Biting was first observed from 2 to 6 min with the results of Blanchard et al . (1975),
after the introduction of the stimuli . Differences which measured the distribution of wounds on
in bite latencies for the stimulus groups were unanaesthetized strangers : in that study less
statistically significant (F = 8 df = 3 and 15, than 10% of lesions were on the head, strongly
P < 0 .01) . Differences in latency to bite were suggesting that unanaesthetized rats are capable
not significant for the dorsal and twisted stimulus of protecting their heads even when severely
groups, nor were differences between the ventral attacked by a dominant male .
and ventral-posterior groups significant . How- The head, however, is the primary attack site
ever, bite latencies were significantly longer for when unlimited access is available. The back is
the ventral group as compared to the twisted clearly another major target : for both the
group (P < 0-05) while the ventral-posterior twisted and the dorsal groups, in which the
group had longer bite latencies than either the head and back were both readily available, the
twisted or the dorsal groups (P < 0 . 01, in each overwhelming majority of bites or obtained
case) . Bite latencies to the ventral group were lesions in areas other than the head, occurred
greater than those to the dorsal group (P < 0 . 05) . to the anterior back. For the ventrally-presented
These differences in the latency to bite are to stimulus rats, in which the dorsal aspect was
some extent attributable to the fact that all six relatively unavailable, only a single correspond-
alpha males in these colonies showed biting ing lesion was found on the ventral anterior
attacks toward dorsal and twisted stimulus surface, the thorax .
rats, while only four of the six showed a biting When both the head and the anterior portions
attack toward the ventral stimulus animal and of the body (both dorsal and ventral) were made
only three of the six toward the ventral-posterior completely unavailable for attack by covering
stimulus rats . Differences in the average number these regions with a transparent box, a dramatic
of bites made at the different stimulus groups shift in lesion site occurred . The rats attacking
failed to reach an acceptable level of statistical these stimuli were faced with a large expanse
significance (F = 3 . 16, df = 3 and 15, 0-10 < of abdomen and anogenital area, with the inner
P > 0-05) . However, the four stimulus groups flanks of the stimulus rat stretched out at the
626 ANIMAL BEHAVIOUR, 25, 3

sides . However, of the 34 wounds found on ventrally presented rats were covered by a clear
such stimulus animals, only four were located Plexiglas box . Since these stimulus rats were
on abdomen or ventral anogenital areas, which completely visible to their attackers it is not
constituted much more than half of the target possible to make the case that the posterior-
surface . Instead, attacking rats tore at the hind- ventral portions alone of an introduced stranger
limbs of the stimulus rats (59% of the observed are sufficient to elicit attack . This point is an
lesions) or burrowed underneath the stimulus important one in clarifying the factors leading
animal to bite the posterior back and the lateral to a reduction of biting attack on the ventral
and caudal surfaces of the posterior limbs (29 surface of the rat .
of wounds) .
These data, then, clearly suggest an ordering Attack on Posterior-Ventrally Exposed Stimulus
of body areas as attack sites, with the head first, Rats
upper back second, and lower back next . If all When the two colony males showing the most
these sites are made relatively unavailable, the rapid and highest amplitude attacks on intruders
flanks may be attacked but not the ventral were presented with ventrally-exposed anaesthe-
surface of the trunk . Further, the number of tized strangers for which the anterior portions of
alpha males showing any biting attack at the the body were covered by an opaque Plexiglas
stimulus rats tends to fall off as the favoured box, both of these alpha males showed rapid
target sites are eliminated. piloerection and biting attack to the introduced
Although these data are consistent it should be stimulus rats . The colony 6 male showed pilo-
noted that the posterior-ventral lesion locus erection in an average of 13 . 11 s and the colony 7
data are based on those three (of six) alpha male in an average of 21 . 85 s . Latency to the
males who attacked the box-covered stimulus first bite averaged 1 . 48 min for the colony 6
group . Since each of these dominant males was male and 1 . 61 min for the colony 7 male . Both
presented with only one rat in the posterior- males spent an average of more than 515 s of
ventral position, statistical analysis of the locus the 600-s test sessions in close contact with the
data becomes very difficult. Another factor stimulus rats and platform . The colony 6 male
which makes interpretation of these data less made an average of 7 . 14 anal sniffs and the
than conclusive is the fact that the posterior- colony 7 male 8 . 71 sniffs during the 10-min

Table II. Percentage of Lesions for Each Stimulus Group, Made at Specific Dorsal or Ventral Sites

Dorsal Loci

Anterior Anterior Posterior Posterior

Stimulus Head limbs back back limbs Anogenital

Dorsal 52 . 70 2 . 70* 13 . 51 * 4 . 05* 1 . 35* 0 . 00*

Twisted 60 . 00* 4 , 00* 14 . 00* 0 . 00 0 .00 0 . 00

Ventral 42 . 85 0 . 00 4 . 76 4 . 76 0 . 00 0 . 00

Ventral posterior 11 . 76 11 .76 5 . 88

Ventral Loci
Anterior Posterior
Stimulus Head limbs Thorax Abdomen limbs Anogenital

Dorsal 25 . 67 0 . 00 0 .00 0 .00 0 .00 0 . 00

Twisted 22 . 00 0.00 0 . 00 0 . 00 0 . 00* 0 .00*

Ventral 42 . 85* 0 . 00* 4 .76* 0 .00* 0 . 00* 0 . 00*

Ventral posterior 2 . 94 58 . 82* 8 . 82*

*Areas which were exposed in each condition .

 BLANCHARD ET AL. : RAT ATTACK AND DEFENCE 627

sessions . It might be noted that attack be- suggest that such sniffs are functional in deter-
haviour of these male rats did not decline mining whether the introduced rat is a stranger .
over the seven stimuli presented in a 2-day If this view is correct it indicates that the
period . same areas which tend not to be bitten during
The distribution of lesions on the introduced conspecific attacks are those which provide
strangers was very similar to that found in information essential to eliciting or releasing
experiment 1 . For both dominant males the the attack reaction . This phenomenon therefore
vast majority (over 95 %) of attacks involved indicates that the anogenital regions do not
the leg . Alpha male 6 also made two lesions on escape biting attack merely because of their in-
the foot of a single stimulus rat and one lesion significance to the attacking rat . Thus, there may
on the stomach of another rat . Alpha male 7 be some biological constraint, or limitation, on
made two lesions on the foot of one stimulus ventral tunk bites in conspecific rat encounters .
rat, bit the stomach of another (one lesion), A more conservative view is that these areas are
and made lesions in the scrotum, genital area, simply the least favoured attack sites for attack-
and stomach of a third . Thus, no lesions were ing conspecifics . However, regardless of which
found outside the thigh and foot areas for six specific interpretation is made, experiments I
of the seven rats presented to alpha male 6 and 2 provide definitive evidence of a gradient
and for five of the seven rats presented to alpha of attack probabilities to different bodily sites
male 7 . In contrast, these two rats made an during conspecific agonistic encounters involving
average of more than 15 lesions in the legs and albino rats . Further, these experiments indicate
feet of each introduced stranger . that this non-random distribution of attack
sites does not depend on accessibility of bite
Statistical analysis of these data fully sup- targets, or on the behaviour of the rat attacked .
ports the notion of a non-random distribution of It might be noted that Scott suggested in 1966
lesions on the stimulus rats . Since the stomach, that attacking rats tend to bite upper back
anogenital and buttock areas are larger than the areas, and Fox (1972) has reported a similar
thighs and feet, direct comparison of the non-random distribution of bites in conspecific
numbers of bites to these different body portions canid encounters .
seemed justified : a sign test was therefore used Given this gradient of bite targets, plus the
to assess the statistical significance of the possibility that the vibrissae may serve as a head-
differences in the number of lesions in these protection device for unanaesthetized defenders
different loci : this test indicated that significant- (Blanchard et al . 1975) it is of considerable
ly more lesions were made to the leg areas of interest to determine if the behaviour of defend-
stimulus rats than on the ventral surface of ing strangers reflects these two potential sources
the trunk (T = 0, N = 7, P < 0.05) for each of protection against biting attack .
animal .
The present finding of high and consistent Alterations of Attack and Defensive Behaviour
lesion rates for the stimulus animals in the After Anaesthetization of Intruder Vibrissae
present study, obtained when the anterior Table III presents mean times for each of the
portions of these rats were covered by an opaque behaviour categories for introduced animals
box, strongly indicates that the sight of a stimulus with vibrissae pads anaesthetized, for non-
rat's head and forward trunk is not necessary anaesthetized control intruders, and for alpha
to elicit attack by a dominant conspecific . males confronted by these two types of stimulus
Any reduction in attack behaviour to the ventral animals .
surface of a stimulus rat's body cannot, there- Attack and defensive patterns . As Table III
fore, be attributed to an inability of this pre- indicates, there were several substantial differ-
paration to elicit attack behaviour . Also, in ences between the attackers' behaviour and the
view of the fact that anogenital sniffs always behaviour of control intruders . The attacking
occurred during the first seconds of confronta- alpha males made reliably more chase and
tion with a new stimulus rat, it is apparent that on-top-of reactions (U = 9, N = 6, P < 0 .05
this area is an attractive stimulus for the in- in both cases) and tended to make more lateral
vestigatory responses of the dominant males . displays . The lateral display differences, how-
Observations of piloerection after sniffing the ever, failed to reach an acceptable level of
anogenital area of strangers, but not of male statistical significance (two of the alpha males
rats from the same colony as the attacker, made no lateral displays, U = 13) . The intro-
6 28 ANIMAL BEHAVIOUR, 25, 3

duced control rats made more flight, on-the- Effects of anaesthetization of defender's vibris-
back, and freezing responses (U = 0, N = 6 sae. Anaesthetization of the vibrissae pads
and 6, P < 0 . 05 in each case) . Boxing responses produced a straightforward pattern of be-
were more common for the control animals, but havioural differences in the introduced strangers .
failed to reliably differentiate the two groups The animals with anaesthetized vibrissae were
(t = 2 . 12, df = 10, 0 .05 < P < 0 . 10) . These sharply less inclined to box but spent much more
data therefore substantiate very nicely the time freezing. All other categories of behaviour
suggestion that agonistic behaviour of the albino were relatively unchanged : in fact, the reduction
rat is very different for attacking and defending of 148 . 55 s in mean boxing time for the anaesthe-
animals. The only behaviour categories here tized vibrissae animals was almost exactly
that did not show a substantial difference for accounted for by an increase of 146 . 82 s in
attacking and defending animals were contact freezing . Both of these differences between
time (which was necessarily the same for alpha anaesthetized and control strangers were statis-
and the stranger, since it was contact time tically reliable (t = 2 .41 and t = 2 . 33, res-
between the two animals) and roll, tumble, pectively, df = 10, P < 0 . 05) . It should be
and biting behaviour. This last reaction could noted that the incidence of the on-the-back
be measured independently for the two animals, posture was very high for both groups of
as alpha would sometimes bite at the stranger defending animals and changes little with an-
while the stranger was involved in some other aesthetization of the vibrissae . However, an-
behaviour (usually on-the-back or boxing), but aesthetization did appear to produce a qualitative
it was frequently a reaction in which both change in the on-the-back posture, since control
animals participated with vigorous movements (but not anaesthetized) rats tended to partly
by each . Finer analysis may ultimately show that orient their head, snout, and vibrissae to face
the qualitative reactions of the attacking and the attacking animal .
defending animals during what is here called a Differences in the behaviour of the attacking
roll, tumble, and bite response are different . colony male to control animals or to animals
However, the present analysis did not differ- with anaesthetized vibrissae appeared to largely
entiate the two . involve three response categories, the roll,
tumble, and bite category and the on-top-of
category, both of which became more frequent
Table III . Behaviours of Dominant Colony Males Con- in response to rats with anaesthetized vibrissae,
fronted by Introduced Rats with Intact (Control) or
Anaesthetized Vibrissae, and Behaviours of the Introduced and the lateral display category, which became
Strange Rats less frequent . Due perhaps to the small number
of alpha males used, none of these differences
Alpha behaviour Stranger behaviour reached an acceptable level of statistical sig-
nificance, and they may therefore be taken only
Anaes- Anaes- as suggestive of a relationship between defender
thetized thetized
stranger Control stranger Control behaviour and attacker behaviour .
z z x z Bites on introduced strangers . Table IV
Contact time (s) 430 . 35 412 . 20 430 . 35 412 .20 presents the number and loci of lesions for
control rats and for rats with anaesthetized
Roll, tumble 45 . 30 14 . 30 17 . 98 12 . 78 vibrissae . These data add considerably to the
and bite analysis proposed of the functions of specific
Chase or flee 4 .08 8 . 38 12 . 73 21 . 18 defensive and aggressive behaviours in the rat,
in that they clearly indicate a differential dis-
Box 1 . 27 14 . 35 1 . 22 149 . 77 tribution of bites on animals in which the vibris-
sae were anaesthetized as compared to control
On-the-back 0 . 00 0 . 00 287 . 63 284 . 62 rats . The head, especially, of anaesthetized rats
Rear immobile 6 .03 10 .21 2 . 53 11 . 63 received more bites than did the corresponding
area of control animals . This difference was
Freeze 0 .00 0 .00 201 . 85 55 . 03 statistically significant (U = 0, N = 6 and 6,
On top 180 . 37 82 . 40 0 . 00 0 . 00
P < 0 . 01) . The anterior back also was bitten
more frequently in these animals, but this
Lateral display 1 .20 20 .28 0 .00 0 .00 increase was not significant. Increases in bites
to these two areas were responsible for the overall
 BLANCHARD ET AL. : RAT ATTACK AND DEFENCE 629

bite increases seen to stimulus rats with an- introduced strangers divided their defensive
aesthetized vibrissae . behaviour evenly between boxing and lying
This pattern of findings, that anaesthetization `on-the-back' .
of the vibrissae produces a sharp decrease in the Intruder boxing . Table V presents major activity
defensive boxing posture, and allows greater categories for the attacking (alpha) rat, while the
numbers of head bites on the anaesthetized rat, defending (intruder) rat was boxing . As this
clearly suggests that the vibrissae normally are table indicates, boxing in the introduced stranger
used in conspecific agonistic encounters, to was associated with very high levels of the lateral
protect the defender's head. Certainly these data display, for the attacking alpha : during two-
do not support contentions, such as that of Thor thirds of the intruder boxing sequences, alpha
et al. (1974) that the vibrissae are necessary for rats were in the lateral display. This finding is
aggression . Instead, the present results are in fully congruent with a view that the lateral
agreement with Scott & Fredericson's (1951) display is a strategy useful in permitting the
suggestion that the boxing seen in `reflexive attacking rat to dart around and bite at target
fighting' tasks, is a defensive, rather than an sites (back and sides) of the intruder .
aggressive behaviour . These results are also This view is further supported by the high
congruent with the notion that other defender and consistent number of intention movements
behaviours may involve attempts to interpose made by alpha while in the lateral display : if the
less-vulnerable areas, such as the ventral trunk, function of the lateral display is to facilitate
between the attacker, and favoured (dorsal) quick movements around and to the back of the
attack sites . defending rat, then such movements should be
Interaction of Attack and Defender Behaviour : a consistent feature of periods in which alpha
Film Analyses rats remain a lateral display. In fact, on average,
one intention movement occurred every 2 s
Film analysis of agonistic sequences indicated during the lateral display .
that the intruder's behaviour largely consisted Although the lateral display was the major
of either boxing or lying on-the-back . These attacker activity shown to boxing intruders,
two defensive behaviours were rather unevenly some instances of intruder boxing involved
distributed in relative frequency for the six maintenance of the alpha, as well as the intruder,
introduced animals (Table V) . Three of the six in a brief episode of the boxing posture . In
spent much more time boxing than lying on-the- fact, for four of the five alpha rats, the lateral
back ; one never boxed, but showed high fre- attack and boxing were the only two activities
quencies of lying on-the-back ; the remaining two seen when the intruder was in the boxing posture .
Table VI indicates the outcome, in terms of
Table IV . Mean Number of Lesions on Rats alpha bites on the intruder, of the different
with Anaesthetized Vibrissae, or, Controls types of alpha behaviour . As this table indicates,
the alpha lateral display was very different from
Anaesthetized Control alpha boxing, in terms of its bite outcome . The
. Range R Range lateral display was a consistent precursor of
biting : five of the six alphas made 20 % or more
Head 7 . 67 1-20 1 . 17 0-6 of their bites from a lateral display position,
Anterior back 12 . 83
while the intruder was in the boxing posture .
4-29 8 . 83 3-23 In contrast, not one single bite was made while
Posterior back 4 . 50 0-10 5 . 50 0-13 alpha was in the boxing posture .
This finding makes it clear that, in response
Limbs 0 . 00 0-0 0 . 67 0-2 to intruder boxing, the lateral display is a suc-
Dorsal anogenital 0 . 00 0-0 0 . 17
cessful strategy, i .e. one which consistently
0-1
results in an opportunity to bite the intruder .
Thorax 1 . 00 0-4 0 . 00 0-0 In striking contrast, if alpha boxing in response
Abdomen 0 . 33 0-1
to intruder boxing, is viewed as an attack
0 . 17 0-1 strategy of the colony alpha, it is clearly a
Anogenital 0 . 00 0-0 0 . 00 0-0 failure . However, if such alpha boxing is viewed
as the outcome of attempts by the intruder to
Total 26 .33 16 . 51 maintain vibrissae-to-vibrissae contact, it is
clear that this outcome is highly successful for
630 ANIMAL BEHAVIOUR, 25, 3

the intruder, who is never bitten so long as it intention movement was the `check', a shift in
can maintain vibrissae contact with alpha . bodily orientation (usually turning or twisting
Table VII presents intruder reactions to alpha to continue a frontal position with reference to
intention movements, while the intruders were the attacker) which continues the status quo .
in the boxing posture, or lying on-the-back . The However, the second most frequent response
most common intruder response to an alpha to alpha intention movements was flight.

Table V. Alpha Behaviour as a Function of Intruder Behaviour : Intruder Boxing

Colony
3 4 5 6 7 8 x
(Intruder : % Time boxing) : (77 . 78) (32 .46) (25-13) (63 . 85) (58-87) (0-00) (43-01)
Alpha behaviour
Total contact time, as % 93 . 91 88 .09 91-58 100-00 100-00 - 94 . 72
intruder box time
Time lateral display 46 . 17 91 . 62 51-23 79 . 40 80 . 24 - 69-73
Time boxing 53 . 82 8 . 38 48 . 77 16 . 28 19-76 29 . 40

Time other 0 . 00 0 . 00 0 .00 4. 33 0 .00 0-87

Intention movements, during

alpha lateral display :
Total number 23 27 13 44 41 - 29 . 60

Interval (s) between alpha 2 . 88 1 . 91 1 . 79 2-22 2 . 22 - 2 . 20

intention movements
Intention movements, during
alpha boxing :
Total number 15 1 8 7 14 - 9 . 00

Interval (s) between alpha 5 . 16 4 . 72 2 . 76 2 . 87 1 . 60 - 3 .42

intention movements

Table VI. Biting Behaviour : Alpha bites on intruders : Antecedent Conditions for Bites and Bite Loci

Colony
3 4 5 6 7 8 9

Total bites : 5 8 20 11 17 13 12-33

Pre-bite behaviour (%)

Intruder flee : alpha chase 80 .00 62 . 50 55 . 00 72 . 73 70-59 46-15 64-49
Intruder box : alpha lateral 20 . 00 25 .00 25-00 27-27 23-50 0 . 00 20 .13
Intruder box : alpha box 0 .00 0 .00 0 .00 0 . 00 0 . 00 0 . 00 0 .00
Intruder on-the-back : alpha on top 0 .00 12 . 50 20 . 00 0 . 00 5-88 53 . 85 15 .37

Bite locus (%)

Head 0 . 00 0 . 00 0 . 00 0 . 00 0 . 00 0 . 00 0 .00
Anterior back 0 . 00 37 . 50 25 . 00 27-27 11 . 76 76-92 29-74
Posterior back 100 . 00 62 .50 55-00 72-73 76 . 47 23-08 64-96
Thorax 0 . 00 0 . 00 0 . 00 0 . 00 0 . 00 0 . 00 0 .00
Abdomen 0 . 00 0 . 00 10 . 00 0 . 00 0 . 00 0-00 1-67
Legs 0 .00 0 . 00 10 . 00 0 . 00 11-76 0-00 3-63
Tail 0 . 00 0 . 00 0 . 00 0 . 00 0 . 00 0 .00 0-00
 BLANCHARD ET AL . : RAT ATTACK AND DEFENCE 631

As Table VII indicates, flight was clearly the although it was maladaptive in the colony
most dangerous response for the intruder in the situation .
present colony situation . Although flight was a Intruder on-the-back. Alpha responses to
relatively infrequent and very short-lasting intruders who were lying on-the-back (Table
response, accounting for only 5 . 6 % of intruder VIII) were consistently different than those made
time, it preceded nearly 65 % of the total bites to boxing intruders. When intruders assumed
given during these agonistic encounters . This this co-called `submission' posture, the alpha
finding is fully congruent with the view that approached closely, even standing directly over
intruder boxing is a strategy which is usually the prone intruder . The alpha orientation per-
successful (especially when it results in a 'trap- pendicular to the long axis of the intruder is here
ping' of the alpha into the boxing posture also) . called on-top-lateral, to suggest that it has a
However, when alpha intention movements common function with the lateral display, al-
break up intruder boxing and produce flight, though in the lateral display alpha moves its
exposing dorsal attack sites, a very high bite body sideways toward the facing intruder,
level results . while in the on-top-lateral behaviour it is
The fact that intruder flight was such a poor alpha which is facing (and on top of) the intruder .
strategy in terms of bite outcomes raises the In both cases, however, such a perpendicular
question of why intruders continue to flee . In orientation of alpha to intruder facilitates
contrast to the alpha rats which had had con- avoidance of the intruder's vibrissae by the
siderable agonistic experience, both in terms of attacking animal . Thus there is a possible
introduced strangers and in terms of the activities functional similarity of lateral display and the
involved in the establishment of colony domi- on-top-lateral posture .
nance hierarchies, the introduced rats were As a comparison of Tables VI and VII in-
naive and had been individually housed for dicates, the on-the-back posture was relatively
virtually all of their adult lives . This factor dangerous for the intruding rat : nearly 15
strongly suggests that the specific strategies of bites received by these animals occurred while
involved in the intruders' efforts at defence were the intruder was on-the-back . Thus, in terms of
not learned during the lifetime of the individual bites per unit time, on-the-back was about as
animals, but instead represent species-typical, dangerous (or, conversely, as successful) as
or pre-programmed defensive responses . The the boxing response . However, both of these
fact that flight (in response to alpha intention intruder responses were much more likely to
movements) continued to occur even though it avoid alpha bites, than was the flight response .
was almost invariably followed by a bite to the This finding, that the on-the-back reaction is
intruder, suggests that the tendency to flee is associated with a bite risk equivalent to that of
poorly controlled by the outcome of flight . the boxing response, also strongly suggests that
Thus, since flight is usually an adaptive response the on-the-back posture does not serve as an
in intraspecific encounters for the rodent it has attack-inhibiting `submission signal' in the rat :
become a very dominant response tendency the attackers continued to press their attack on

Table VII. Intruder Reactions to Alpha Intention Movements, While Intruder Boxing, and On-The-Back

Frequency of Colony
Intruder Alpha intruder
behaviour behaviour reactions 3 4 5 6 7 8 z
Box Lateral display with intention movements Check 21 21 7 33 28 22.00
Flight 2 5 3 10 4 920
No defence 1 1 3 1 9 - 3 .00
Box Boxing with intention movements Check 13 1 8 7 14 - 8 . 60
Flight 2 0 0 0 14 3 .20
No defence 0 0 0 0 0 - 0 .00
On the On-top with intention movements Check - 3 6 1 5 8 4.60
back Flight - 0 2 0 1 0 0.60
No defence - 20 17 0 1 32 14 . 00
6 32 ANIMAL BEHAVIOUR, 25, 3

strangers in this posture, such that all strangers outside these two loci . Additionally, the total
who spent as much as 1 s in the on-the-back absence of bites to the head areas of defenders
posture, were bitten in this posture . This is clearly is congruent with the view that the vibrissae
incompatible with a view that the on-the-back provide head protection for defenders .
posture serves as a signal of submission or defeat One final area of evidence is very relevant to
which inhibits biting attack . the interpretation of specific defensive be-
Thus, these data are fully supportive of a view haviours in terms of utilization of constraints
that both attacker and defender behaviour of on attack loci : the most common (96 . 34%)
rats in a conspecific agonistic encounter are immediate defender response to being bitten
based on strategies which have proved successful was a turn, generally toward the attacker . Over
in the evolutionary history of the species . 70% of defenders then assumed the boxing
Specifically, these data indicate that intruding posture, or lay on-the-back . Since the defending
rats can escape being bitten by colony alpha rats almost never bit back at alpha, the finding
males in so far as they can interpose vibrissae that they turned toward the attacker after being
and ventral surfaces between the attacker and bitten is understandable only in terms of a view
their own (vulnerable) head and dorsal areas that this frontal orientation, with vibrissae
boxing or lying on-the-back are therefore and ventral thorax proximal to the intruder,
relatively adaptive behaviours for non-dominant provides protection against biting . Thus this
rats . Conversely, any attacker behaviour which finding provides additional support for the view
permits access to these vulnerable surfaces, that constraints on biting attack determine
tends to result in bites . This factor is presum- defender as well as attacker behaviour .
ably involved in the evolution of the lateral
display, the `on-top-lateral' posture, and in- Discussion
tention movements in general, as dominant The four experiments reported here are each
behaviours of animals at the top of group hier- consistent with, and each adds evidence to, the
archies . hypothesis that both attacker and defender
Two final sets of findings are relevant to the behaviours during conspecific agonistic en-
question of the function of defender behaviour counters in the albino rat reflect strategies based
with reference to `protected' areas of the body . on biological constraints on aggression in this
First, Table VI provides consistent evidence of species .
the continued inhibition of biting attack on the Specifically, the major finding of experiment 1
ventral surface of defending rats : over 94 was that, even with unlimited access to anaesthe-
of the filmed bites were made on the centre and tized rats, dominant colony males do not bite
posterior back of the introduced rats . In fact, the thorax, abdomen, or genitals . Instead,
only two of the six alpha rats made any bites biting attack was largely confined to the head

Table VIII. Alpha Behaviour as a Function of Intruder Behaviour : On-The-Back

Colony

3 4 5 6 7 8 2

(Intruder : %Time On-Back) : (0 . 00) (35 . 56) (31 . 08) (0 . 28) (7 . 57) (68 . 25) (23 . 79)
Alpha behaviour
Total contact time as % intruder - 100 . 00 100 . 00 100 . 00 100 . 00 100 . 00 100 .00
On-the-back time
Time On-top-Lateral - 98 . 57 92 . 74 0 . 00 37 . 26 92 . 31 64 . 14
Time On-top-Parallel 1 .43 7 . 27 100 .00 62 . 73 7 . 69 35 . 82
Time other - 0 .00 0 .00 0 .00 0 . 00 0 . 00 0 . 00

Intention movements, while alpha

on-top parallel and lateral, combined
Total number - 23 25 1 7 40 19 . 20
Interval (s) between alpha - 3 .05 2 .45 0 . 56 2 . 09 3 . 36 2 . 30
intention movements
 BLANCHARD ET AL . : RAT ATTACK AND DEFENCE

and upper back of intruders . Even when only Specifically, experiment 3 suggested the fol-
the ventral and posterior areas of intruders lowing analysis of agonistic behaviours : attack-
were exposed, the ventral trunk remained un- ing rats aim their bites at head and back areas ;
bitten . Additionally, experiments I and 2 the defender, in assuming the boxing stance,
consistently indicated that alpha rats continue conceals its back areas by interposing thorax,
to attack intruders with only the ventral and abdomen, and genitals between its vulnerable
posterior surface exposed, burrowing under to areas and the attacker, and also protects its
bite, or biting the flanks. This strongly indicates head and shoulders by attempting to engage the
that thorax, abdomen, and genitals served as a attacker in vibrissae-contact . The attacker
sufficient stimulus to elicit attack . The lack of evades vibrissae-contact by lowering its head
attack then, is suggestive of some biological and turning `broadside' to the defender, crowd-
constraint on bites to these areas . ing against the defender's body in a lateral
This logical point might perhaps be re- display. This crowding tactic may cause the
emphasized : experiments I and 2 do not prove defender to break away and flee, giving the
conclusively that there is a constraint on biting attacker a clear shot at the defender's back,
attack to thorax, abdomen, and genitals of or, the attacker may lunge forward and around
conspecifics in the albino rat . It is possible to to bite quickly at the back . However, if the
maintain the alternative position, that there is defender does not flee, but pivots to continue
simply a hierarchy of target sites eliciting biting to face the attacker (check), then the attacker's
attack in conspecific fighting with the ventral intention movement will not result in an actual
trunk very low on this hierarchy . This alterna- bite . Under severe attack, the defender will
tive formulation does leave open the question of assume an on-the-back posture which provides
why alpha rats piloerect when presented with even better protection for the back, but which
the ventral posterior portions of an anaesthetized considerably limits the animal's mobility . Even
stranger, and then bite savagely at the hind- in this position, however, the defender often
limbs of these animals, still failing to bite the pivots to track or check a circling attacker,
ventral trunk . The exact mechanism, then, keeping its thorax, stomach, genitals, and ex-
underlying the present nonrandom distribution tended vibrissae interposed between the attacker
of biting attack on intruders, remains unsettled . and its own vulnerable areas.
However, the precise nature of this mechan- Each of these analyses is supported by the
ism is not particularly important to the major findings of experiment 4, which provided in-
hypothesis underlying the present investigations . formation on attacker responses to boxing, or to
Regardless of the specific reason why attack lying on-the-back by the defender . The bite data
on the ventral trunk is not made, it is clear that of experiment 4 also made it clear that the lateral
this area does not get bitten . Thus, the inter- display is a successful (i.e . likely to result in an
position of thorax, abdomen, and genitals opportunity to bite) response to the defender
between an attacker and those defender sites boxing posture, while the on-top---lateral be-
which do get bitten is a procedure likely to haviour is equally successful in response to a
result in fewer bites . prone defender.
Experiment 3 suggested an additional con- The major conception to emerge from these
straint on agonistic behaviour, that attacking findings is, therefore, the view that the specific
rats may be prevented from directly pressing defensive acts of rats are strategies which enable
their attack to the head and upper back of the a defender to utilize limits on its opponents
defender, by vibrissae-contact with the defender . attack behaviours . Similarly, specific aggressive
This constraint had the effect of altering the responses, such as the lateral display, may be
distribution of bite wounds to anaesthetized viewed as strategies permitting the attacker to
and unanaesthetized strangers, since the an- circumvent specific defensive manoeuvres of its
aesthetized animals could not use their vibrissae conspecific opponent . It is this mutual depen-
to defend the head and upper back . Experiment dence of attacker and defender behaviour which
3 also detailed some of the specific defensive gives conspecific agonistic interactions in the
acts by which a defending rat may maximize rat their stereotyped 'dance-like' character .
the effectiveness of these constraints on the In terms of the possible evolutionary develop-
attacker's behaviour, as well as suggesting that ment of these factors, the present view would
some aggressive acts may enable the attacker to suggest that the limits on biting attack underlie
counter specific defensive behaviours . the specific defensive strategies observed, rather
6 34 ANIMAL BEHAVIOUR, 25, 3

than the opposite direction of influence . That suggest that a very substantial agreement may be
is, a view that the reluctance of dominant obtained between the agonistic reactions of the
animals to bite the thorax, stomach, and genitals present subject strain and those of a number of
of conspecifics is due to the fact that these are wild rat species (compare these behaviours with
presented when the defending animal is in a those reported by : Barnett 1963 ; Eibl-Eibesfeldt
submission posture, simply will not work : when 1961 ; Ewer 1971). Thus available evidence
strangers were presented in a submissive ventral suggests that some limitation on attack to thorax,
or posterior-ventral position, the dominant abdomen, and genitals of conspecifics may be a
animals bit their heads, or burrowed under to feature of agonistic interactions in all rat species,
bite the back or bite the limbs, but failed to bite a feature which greatly influences the specific
the ventral trunk areas (experiments 1 and 2) . reactions of both attacking and defending
Also experiment 4 indicated that substantial animals in this genus .
numbers of bites were made to unanaesthetized
strangers while they were in the so-called REFERENCES
submissive posture . Such findings make it clear
Alberts, J . R . & Galef, B. G ., Jr. 1973 . Olfactory cues
that an on-the-back posture does not facilitate and movement : Stimuli mediating intraspecific
defence by inhibiting or reducing the overall aggression in the wild Norway ray . J. comp .
attack propensities of the dominant animal . physiol . Psycho!., 85, 233-242 .
These data therefore strongly suggest that it is Barnett, S. A . 1963 . The Rat : A Study in Behaviour.
the biting constraint which constitutes the Chicago : Aldine Publishing Company .
primary determinant of the form of conspecific Blanchard, R. J., Fukunaga, K . K., Blanchard, D. C . &
Kelley, M . J . 1975 . Conspecific aggression in the
defensive or aggressive acts in the rat. This view laboratory rat. J. comp. physiol. Psychol ., 89,
of aggressive and defensive reactions as strategies 1204-1209 .
may be contrasted with the more traditional Eibl-Eibesfeldt, I . 1961 . The fighting behavior of
description (cf. Grant & Mackintosh, 1963) of animals . Scient. Am., 205, 112-122 .
such activities as boxing, lying on-the-back and Ewer, R . F . 1971 . The biology and behaviour of a free-
the lateral display, as displays or signals given living population of black rats (Rattus rattus) .
Anim. Behav . Monogr ., 4, 127-174 .
by one animal to the other, to communicate Fox, M . W . 1972 . The Behavior of Wolves, Dogs and
threat, submission, defeat, or other relevant Related Canids. New York : Harper & Row .
characteristics of the signalling animal . While Grant, E. C . & Mackintosh, J . H. 1963 . A comparison
of the social postures of common laboratory
it is quite probable that purely communicatory rodents . Behaviour, 21, 260-281 .
signals or displays may play some role in Scott, J. P. 1966 . Agonistic behavior of mice and rats :
agonistic behaviour, the present results suggest A review . Am. Zool., 6, 683-701 .
that it is not necessary to ascribe a signal Scott, J . P. & Fredericson, E . 1951 . The causes of fighting
in mice and rats . Physiol. Zool., 24, 273-309 .
function to these specific behaviours, in order to Telle, H. J . 1966 . Beitrag zur Kenntnis der Verhaltens-
explain their value in the agonistic interactions weise von Ratten, vergleichend dargestellt bei,
of the rat. Rattus norvegicus and Rattus rattus. Z. Angewandte
Finally, it should be noted that the generality Zoologie, 53, 129-196 . (Available in English as
Technical Translation 1608, Translation Section,
of such constraints on attack remain to be National Science Library, National Research
established. However, Ewer (1971) has suggested Council of Canada, Ottawa, Ontario, Canada.)
that attacking wild Rattus rattus make pre- Thor, D. H ., Ghisselli, W. B . & Lambelet, D . C. 1974.
dominantly back bites on conspecifics, while Sensory control of shock-elicited fighting in rats.
Physiol. Behav ., 13, 683-686.
Scott (1966) noted a preponderance of back
bites among albino rats. Also, the present study (Received 4 November 1975 ; revised 12 March 1976 ;
and that of Blanchard et al . (1975) strongly MS. number : A1783)