United Kingdom House of Lords Decisions

OPINIONS

OF THE LORDS OF APPEAL

FOR JUDGMENT IN THE CAUSE

Mirvahedy (FC) (Respondent) v. Henley and another (Appellants)

ON

THURSDAY 20 MARCH 2003

The Appellate Committee comprised:

Lord Nicholls of Birkenhead

Lord Slynn of Hadley

Lord Hobhouse of Woodborough

Lord Scott of Foscote

Lord Walker of Gestingthorpe

HOUSE OF LORDS

OPINIONS OF THE LORDS OF APPEAL FOR JUDGMENT

IN THE CAUSE

Mirvahedy (FC) (Respondent) v. Henley and another (Appellants)

[2003] UKHL 16

Lord Nicholls of Birkenhead

My Lords,

    1. Shortly after midnight on the night of 28-29 August 1996 Hossein Mirvahedy was driving home from his work as manager of a hotel in Devon. He was driving along a dual carriageway stretch of the A380 from Torquay to Exeter. His car came into collision with a horse when it ran across the road and crashed into the car. He suffered serious personal injuries.

    2. The horse belonged to Andrew and Susan Henley. It had escaped from the field where it was kept. Dr and Mrs Henley lived about a mile from where the accident occurred. In an adjacent field they kept three horses, of which the horse involved in the accident was one. On the night of the accident all three horses stampeded out of a corner of their field. They pushed over an electric wire fence and a surrounding wooden fence, and then trampled through a strip of tall bracken and vegetation. Something seems to have frightened them very badly, but nobody knows what it was. The horses fled 300 yards up a track and then for a distance of almost a mile along a minor road before reaching the busy A380 road.

    3. Such behaviour is usual in horses when sufficiently alarmed by a threat. They attempt to flee, ignoring obstacles in their way, and are apt to continue in their flight for a considerable distance, even beyond the point where the perceived threat was detectable.

    4. Mr Mirvahedy brought a claim against Dr and Mrs Henley as keepers of the horse. He based his claim in negligence. He said Dr and Mrs Henley had not fenced the field properly. The judge, Judge O'Malley sitting in the Exeter County Court, rejected this claim. No appeal was pursued from this decision. Mr Mirvahedy also advanced a claim under section 2 of the Animals Act 1971. He asserted that, even if they were not at fault, Dr and Mrs Henley were liable for the damage caused by their runaway horse. Under the Animals Act they were strictly liable. They were liable independently of fault. That claim, too, failed before the judge. It succeeded on appeal to the Court of Appeal: see [2001] EWCA Civ 1749, [2002] 2 WLR 566. The court comprised Dame Elizabeth Butler-Sloss P and Hale and Keene LJJ. Dr and Mrs Henley then appealed to your Lordships' House.

    5. The appeal raises one question: is the keeper of an animal such as a horse strictly liable for damage caused by the animal when the animal's behaviour in the circumstances was in no way abnormal for an animal of the species in those circumstances?

    6. Lest there be any misunderstanding one point should be clarified at the outset. Considered as a matter of social policy, there are arguments in favour of answering this question yes, and arguments in favour of answering no. It may be said that the loss should fall on the person who chooses to keep an animal which is known to be dangerous in some circumstances. He is aware of the risks involved, and he should bear the risks. On the other hand, it can be said that, negligence apart, everyone must take the risks associated with the ordinary characteristics of animals commonly kept in this country. These risks are part of the normal give and take of life in this country.

    7. These considerations, and other arguments of this nature, are matters for Parliament. They are not matters for this House acting in its judicial capacity. It is not for the courts to form a view on which of these arguments seems the more weighty when Parliament has already carried out this exercise. Parliament must be taken to have weighed the various factors, and balanced the conflicting interests of those who keep animals and those who are injured by them, when enacting the Animals Act 1971. The answer to the question I have posed lies in interpreting the provisions of this Act, and in particular section 2(2), in accordance with established principles of statutory interpretation.

    8. The common law concerning liability for animals was notoriously intricate and complicated. How the common law would have answered the question raised by this appeal is not altogether clear. The common law may have drawn a distinction between a domestic animal which, contrary to the nature of its species, has a propensity to attack (a 'vicious' propensity), and a domestic animal which, without a propensity to attack, has a special propensity to cause damage. Strict liability, under the old 'scienter' principle, may have been applicable only in the former case: see the discussion in the report of the Law Commission on Civil Liability for Animals (1967) (Law Com no 13), paragraph 6, page 7, and the seemingly differing views of Willmer LJ and Diplock LJ in Fitzgerald v E D and A D Cooke Bourne (Farms) Ltd [1964] 1 QB 249, 258-259, and 270.

    9. The purpose of the Animals Act 1971 was to simplify the law. Sections 1 to 6 of the Act made new provision regarding strict liability for damage done by animals. They replace the old rules of the common law. Section 2 contains provisions relating to liability for damage done by dangerous animals. Unfortunately the language of section 2(2) is itself opaque. In this instance the parliamentary draftsman's zeal for brevity has led to obscurity. Over the years section 2(2) has attracted much judicial obloquy.

    10. Section 2 places all animals into one or other of two categories, according to their species. Animals either belong to a dangerous species, or they do not. The circumstances in which the keeper of an animal is liable for damage caused by his animal depend upon the category to which the animal belongs.

    11. A dangerous species of animal is a species which meets two requirements, set out in section 6(2). A species can include a sub-species or a variety: see section 11. The first requirement (a) is that the species is not commonly domesticated in the British Islands. The second requirement (b) is that fully grown animals of the species 'normally have such characteristics that they are likely, unless restrained, to cause severe damage or that any damage they may cause is likely to be severe'. In short, they are dangerous animals.

    12. A tiger satisfies both requirements. It is not commonly domesticated in this country, and it is dangerous. A horse does not satisfy the first requirement. Unlike tigers, horses are commonly domesticated here. So tigers, satisfying both requirements, are a dangerous species of animals. Horses, which do not satisfy the two requirements, are not.

    13. Section 2(1) imposes upon the keeper of an animal of a dangerous species strict liability for any damage caused by the animal:

'(1) Where any damage is caused by an animal which belongs to a dangerous species, any person who is a keeper of the animal is liable for the damage, except as otherwise provided by this Act.'

    If you choose to keep a dangerous animal not commonly domesticated in this country, you are liable for damage done by the animal. It matters not that you take every precaution to prevent the animal escaping. You may not realise that the animal is dangerous. Liability is independent of fault. Liability is independent of knowledge of the animal's dangerous characteristics. You are liable, subject only to certain defences of general application set out in section 5. These defences apply where the damage was due wholly to the fault of the claimant, or where the claimant voluntarily accepted the risk of damage or was a trespasser.

    14. Section 2(2) established a different regime where damage is caused by an animal not belonging to a dangerous species. This subsection applies, therefore, to all species of animals commonly domesticated here. It includes horses. The material part of section 2(2) provides:

'(2) Where damage is caused by an animal which does not belong to a dangerous species, a keeper of the animal is liable for the damage, except as otherwise provided by this Act, if-

(a) the damage is of a kind which the animal, unless restrained, was likely to cause or which, if caused by the animal, was likely to be severe; and

(b) the likelihood of the damage or of its being severe was due to characteristics of the animal which are not normally found in animals of the same species or are not normally so found except at particular times or in particular circumstances; and

(c) those characteristics were known to that keeper …'

    15. In the present case nothing turns on requirement (a). It is accepted that this pre-condition of liability is satisfied. Similarly with requirement (c): the judge found that this requirement was satisfied in this case, and his finding has not been challenged.

    16. The crucial requirement is requirement (b). Requirement (b) is concerned, in short, with the source of the animal's dangerousness. If requirement (b) is to be met, the dangerousness of the animal, as described in requirement (a), must be attributable to the animal having characteristics falling within one or other of two classes. The first limb of paragraph (b) identifies one class. The animal must have characteristics 'which are not normally found in animals of the same species'. The second limb of paragraph (b) identifies the other class of qualifying characteristics. The animal must have characteristics which are not normally found in animals of the same species 'except at particular times or in particular circumstances'.

    17. Both these classes, it can be noticed at once, are described in terms of abnormality. The first class is that the particular animal has characteristics not normally found in animals of the same species. Unless the relevant characteristic of the animal which caused the damage satisfies this test of abnormality, the case does not fall within the first class. Likewise, the case does not fall within the second class unless the relevant characteristic of the animal is one which, except at particular times or in particular circumstances, is abnormal. The characteristic of the particular animal must be a characteristic which, save on particular occasions, is not a characteristic of animals of the same species.

    18. Thus, the first class embraces a case where animals of the species are normally docile but the particular animal is not. In such a case requirement (b) is met. However, there are many species of animals which are normally docile but which, in certain circumstances or at particular times, behave differently, even dangerously. Dogs are not normally prone to bite all and sundry. But a dog guarding its territory, or a bitch with a litter whose pups are being threatened, may well be vicious. The second class is directed at this type of case. A dog which is prone to bite is likely to fall within the second class. A dog with a general propensity to bite has a characteristic which, save in particular circumstances, is not normally found in dogs. Such a dog has an abnormal characteristic. In such a case requirement (b) is satisfied.

    19. Thus far there is no difficulty. But what of the case where the dog which attacks and bites is at the time acting as a guard dog or is a bitch with pups? Such an animal is behaving dangerously but it is doing so in a manner characteristic of its species in the circumstances. Does such a case also fall within the second class of cases? On this there has been a difference of judicial opinion. This difference of view exists also in your Lordships' House on the instant appeal.

    20. Some judges have held that such a case is within the second class of cases. It falls within the literal language of the statute. The likelihood that the guard dog or the bitch with pups will bite is due to a characteristic of the particular animal which is not normally found in members of the species except in the particular circumstances of guarding territory or protecting pups. To that extent the animal's behaviour was a departure from the normal behaviour of animals of the same species.

    21. I shall call this interpretation of the second limb of section 2(2)(b) 'the Cummings interpretation'. This interpretation was adopted in Cummings v Granger [1977] QB 397. On this interpretation, Mr Mirvahedy's claim succeeds. Horses do not normally behave as did the horses of Dr and Mrs Henley during the night of 28 August 1996. They do so only in particular circumstances, namely, when seriously frightened.

    22. Other judges have said that this type of case is outside the second class of cases identified in paragraph (b). The second limb of section 2(2)(b) does not treat as abnormal behaviour which is characteristic of the species in the circumstances in which it occurred. The object of the second limb is to provide that characteristics not normally found in the species do not cease to be abnormal because, in certain circumstances or at certain times, all animals of the species behave in that way. A dog prone to attack all strangers has an abnormal characteristic. It is an abnormal characteristic even though in some circumstances all dogs are liable to attack strangers.

    23. I shall call this 'the Breeden interpretation'. This interpretation was favoured in the unreported decision of the Court of Appeal in Breeden v Lampard (21 March 1985). On this construction of the second limb of section 2(2)(b), Mr Mirvahedy's claim fails. The horse which escaped from the field and collided with his car was not behaving differently from the way any normal horse would have behaved in the circumstances.

    24. I turn to the court decisions where this point has been considered. There are four relevant decisions, each of the Court of Appeal. Two adopted one interpretation, two favoured the other. In Cummings v Granger [1977] QB 397 an untrained Alsatian was turned loose in a scrap-yard to deter intruders. The dog seriously injured the plaintiff who had entered the yard. The Court of Appeal, comprising Lord Denning MR and Ormrod and Bridge LJJ, held that the requirements of section 2(2) were satisfied but that the defendant was entitled to rely upon the trespasser defence provided by section 5. The dog had characteristics not normally found in Alsatian dogs except in circumstances where they are used as guard dogs. These were 'particular circumstances' within section 2(2)(b).

    25. Next is the unreported case of Breeden v Lampard (21 March 1985). The court comprised Oliver and Lloyd LJJ and Sir George Waller. A riding accident occurred at a cubbing meet. The plaintiff's leg was injured when the defendant's horse kicked out. A claim was advanced under section 2. This horse, like any horse, was liable to kick out when approached too closely, or too quickly, from behind. This, it was said, brought the case within the second class of cases which satisfy requirement (b). The claim under section 2 failed, principally on the ground that even if requirement (b) was satisfied, requirement (c) was not. Lloyd LJ, however, went on to consider the ambit of requirement (b). He preferred the Breeden interpretation, as I have described it:

'If liability is based on the possession of some abnormal characteristic known to the owner, then I cannot see any sense in imposing liability when the animal is behaving in a perfectly normal way for all animals of that species in those circumstances, even though it would not be normal for those animals to behave in that way in other circumstances, for example, a bitch with pups or a horse kicking out when approached too suddenly, or too closely, from behind.'

    26. In his view the second limb of requirement (b) is 'refining what is meant by abnormality, not imposing a head of liability contrary to the main thrust of section 2(2)(b)'. Oliver LJ said he could not believe Parliament intended to impose liability for what was essentially normal behaviour in all animals of the species. The attention of the court seems not to have been drawn to the decision in Cummings v Granger.

    27. The third case is the decision of the Court of Appeal, comprising Slade, Nourse and Stuart-Smith LJJ, in Curtis v Betts [1990] 1 WLR 459. A bull mastiff attacked and injured a young girl as she approached the dog in the street as it was being transferred to a Land Rover in which it was regularly transported. The court held that the keeper of the dog was liable under section 2(2). The court followed Cummings v Granger. This dog, like other bull mastiffs, tended to react fiercely when defending what it regarded as its own territory. But the mere fact that a particular animal shares its potentially dangerous characteristics with other animals of the same species will not preclude the satisfaction of requirement (b) if on the facts the likelihood of damage was attributable to characteristics normally found in animals of the species at times or in circumstances corresponding to those in which the damage actually occurred: see Slade LJ, at page 464. Breeden v Lampard seems not to have been drawn to the attention of the court.

    28. Finally there is the decision in Gloster v Chief Constable of Greater Manchester Police [2000] PIQR P114. The Court of Appeal comprised Pill and Hale LJJ. The plaintiff was a police officer. While carrying out his duties he was bitten by a police dog, an Alsatian, which had been trained to be aggressive when working. The claim failed, largely on the ground that on the particular facts the damage was not caused by the relevant characteristic of the dog. Pill LJ considered the question of interpretation now in issue. He preferred the Breeden interpretation.

    29. In the County Court the judge, while expressing a preference for the dicta in Breeden v Lampard and Gloster v Chief Constable of Greater Manchester, rightly considered he was bound by the decisions in Cummings v Granger and Curtis v Betts. He held, however, that the plaintiff failed on the issue of causation. The Court of Appeal reversed the judge on the latter issue. The accident was caused by the way these horses behaved once they had been terrified: see Hale LJ at [2002] 2 WLR pages 569-571, paragraphs 9 to 17.

    30. In the Court of Appeal Hale LJ held that the proper interpretation of the second limb of requirement (b) is the Cummings interpretation. The words mean what they say. There is nothing in the pre-enactment material to suggest that anything different was intended. In reaching her conclusion Hale LJ placed considerable reliance on the recommendations of the report of the Law Commission on Civil Liability for Animals (1967) (Law Com no 13), which preceded and led to the Act, and on observations of ministers in the course of the legislative passage of the Animals Bill. Dame Elizabeth Butler-Sloss P and Keene LJ agreed with Hale LJ.

    31. In common with all other judges who have had to wrestle with this question, I have found that the tortuous language of section 2(2)(b) renders its intended meaning peculiarly difficult to ascertain. I readily acknowledge that my mind has fluctuated between the two interpretations.

    32. The starting point is to seek to identify the purpose of requirement (b). Stated in general terms the function of requirement (b) is not in doubt. The purpose of this paragraph is to limit the circumstances in which there will be strict liability for damage caused by an animal having the dangerous characteristics described in requirement (a). Possession of such characteristics by an animal (requirement (a)), together with the keeper's awareness that the animal has these characteristics (requirement (c)), is not enough. Meeting requirements (a) and (c) will not suffice. Something more is needed before strict liability arises.

    33. That this is the purpose of requirement (b) is self-evident. Requirement (b) is a pre-condition of liability additional to requirements (a) and (c). That this is the purpose of paragraph (b) is also confirmed by the background to the legislation.

    34. An important part of the background is the Law Commission report, already mentioned. The rationale of strict liability under section 2 can be found in paragraph 17 of this report. If there is to be strict liability for animals of dangerous species, then an animal not within this category should also give rise to strict liability if damage results from dangerous characteristics of the particular animal which are known to its keeper. The keeper of an animal is equally the creator of a special risk if he knowingly keeps a savage Alsatian as if he keeps a tiger.

    35. From this base the Law Commission, in paragraph 18, then took two further steps. The first step was that liability should not be precluded by the fact that a particular animal belonging to a non-dangerous species shared its dangerous characteristic with other animals within the species at certain times of the year or in special conditions. This step, I interpose, is unexceptionable. The Law Commission then gave as an illustration an instance which is, in fact, a further step. The illustration goes beyond the statement of principle which it is intended to exemplify:

'If the keeper of a bitch with a litter knows that it is prone to bite strangers, then even if this is a common characteristic of bitches at such a time, we think that the keeper should be strictly liable'

    36. This illustration is a further step because, unlike the statement of principle, this illustration is of a case where the particular animal was doing no more than behave in a manner characteristic of the species in the circumstances. The Law Commission inferred that strict liability in these circumstances was in line with the common law as stated in Barnes v Lucille Ltd (1907) 96 LT 680. In that case Darling J said, at page 681:

'I do not think … that in order to make the owner of a dog liable that the dog must be always and invariably ferocious. If the owner knows that at certain periods the dog is ferocious, then he has knowledge that at those times the dog is of such a character that he ought to take care of it. If a man knows that a bitch which is ordinarily amiable is ferocious when she has pups, and people go near her, I think he has knowledge that at such times she is of a ferocious character.'

    37. Consistently with this thinking, the Law Commission appended to its report a draft Bill which included in clause 2(2) requirements (a) and (c), as they now are in the Act, but not requirement (b). Clause 2(2) reads:

'Where damage of any kind is caused by an animal which does not belong to a dangerous species, and -

(a) the animal has such characteristics that it is likely, unless restrained, to cause damage of that kind or that any damage of that kind that it may cause is likely to be severe; and

(b) those characteristics are known or treated as known to a person who is keeper of that animal; that person is liable for the damage, except as otherwise provided by this Act.'

    38. Thus, under this subclause abnormality would not be a necessary prerequisite of strict liability. If an animal has dangerous characteristics in respect of a certain kind of damage, and its keeper is aware of this, he is liable if the animal causes damage of that kind. This draft subclause conforms with the Law Commission's recommendations on this point, as set out in paragraphs 18(ii) and 91 (iv) of its report, pages 12-13 and 41.

    39. When first introduced into Parliament in 1969 the Animals Bill was, in the relevant respects, in the same form as the Law Commission's draft Bill. But in the course of its legislative passage there were criticisms of the wording of clause 2(2) of the Bill. The net of strict liability was being cast too widely. Even quite normal behaviour on the part of an animal might cause damage and give rise to strict liability. There was thought to be a risk that, as drafted, a keeper of an animal with a dangerous characteristic could be strictly liable for damage caused by the animal even though the damage was not attributable to the animal's dangerous characteristic. The keeper of a dog with a known propensity to bite could be strictly liable for damage caused if the dog barked and this happened to startle someone.

    40. The Bill lapsed when Parliament was dissolved for the 1970 general election. When the Bill was re-introduced clause 2(2) had been reformulated and what is now requirement (b) had been added. One purpose of the amendments was to confine the scope of section 2(2) to cases where the damage suffered was of the kind falling within requirement (a). So much is clear. Where an animal has dangerous propensities, the keeper is to be liable only for such damage as is due to those propensities.

    41. This purpose furnishes an explanation for the rewording of requirement (a). But it does not provide guidance on the purpose of requirement (b). As drafted, requirement (b) is apt to exclude cases where an animal has a dangerous characteristic as described in requirement (a) but that characteristic is normally found at all times in animals of that species. This is the heart of the problem. The difficulty lies in identifying the type of dangerous characteristic which will satisfy this formula, and thus exclude the keeper from strict liability under section 2(2). Section 2(2)(b) was intended to have some content. The problem is to identify that content.

    42. Neither the Cummings interpretation nor the Breeden interpretation provides a compellingly clear solution to this problem. The principal difficulty with the Cummings interpretation is that it seems to leave section 2(2)(b) with very limited content. This point was well summarised by Judge O'Malley in the present case:

'It is very hard to contemplate or define the characteristics that are not normally found in animals "except at particular times or in particular circumstances". I am concerned at the generalness of words which are expressed as a limitation as to time and circumstance but which can be applied to any case and are therefore no limitation at all. … If the [Cummings] construction … is correct the claimant must succeed in establishing this particular criterion in every case. Either the animal is proved to be an abnormal animal or to have abnormal characteristics or it has normal characteristics upon which the claimant can rely in the particular circumstances of the instant case. For, as it seems to me, all times and all circumstances can be said to be "particular". One can always find particularity attaching to any time or to any circumstance.'

    43. In other words, if the tendency of a horse to bolt when sufficiently alarmed is to be regarded as a normal characteristic of horses 'in particular circumstances' and, hence, a horse with this characteristic will meet requirement (b), it is not easy to conceive of circumstances where dangerous behaviour which is characteristic of a species will not satisfy requirement (b). A normal but dangerous characteristic of a species will usually be identifiable by reference to particular times or particular circumstances. Thus the Cummings interpretation means that requirement (b) will be met in most cases where damage was caused by dangerous behaviour as described in requirement (a). Requirement (b) will be satisfied whenever the animal's conduct was not characteristic of the species in the particular circumstances. Requirement (b) will also be satisfied when the animal's behaviour was characteristic of the species in those circumstances.

    44. This is a cogent argument. Ultimately, despite this argument, on balance I prefer the Cummings interpretation of section 2(2)(b), for a combination of reasons. First, this interpretation accords more easily and naturally with the statutory language. Damage caused by an attack by a newly-calved cow or a dog on guard duty fits readily into the description of damage due to characteristics of a cow or a dog which are not normally found in cows or dogs except in particular circumstances. That is not so with the Breeden interpretation. The Breeden interpretation has the effect that these examples would fall outside both limbs of paragraph (b). This result makes sense only on the supposition that, by the references to abnormal characteristics in section 2(2)(b) (characteristics 'not normally found'), Parliament intended that strict liability should never arise if the animal's conduct was normal for the species in the circumstances in which it occurred. But the language of the paragraph provides no substantial support for this supposition.

    45. Secondly, the Breeden interpretation would depart radically from the legislative scheme recommended by the Law Commission. There is no evidence that any such departure was intended. Indeed, far from such a departure being intended, the wording of clause 2(2)(b) of the reformulated Animals Bill, subsequently enacted as section 2(2)(b) of the Animals Act, was plainly drawn from, and closely followed, the language of paragraphs 18(ii) and 91(iv) of the Law Commission's report.

    46. Thirdly, the 'lack of content' argument levelled against the Cummings interpretation cannot be pressed too far. The Cummings interpretation does not empty requirement (b) of all content. Some forms of accidental damage are instances where this requirement could operate. Take a large and heavy domestic animal such as a mature cow. There is a real risk that if a cow happens to stumble and fall onto someone, any damage suffered will be severe. This would satisfy requirement (a). But a cow's dangerousness in this regard may not fall within requirement (b). This dangerousness is due to a characteristic normally found in all cows at all times. The dangerousness results from their very size and weight. It is not due to a characteristic not normally found in cows 'except at particular times or in particular circumstances'.

    47. For these reasons I agree with the interpretation of section 2(2)(b) adopted in Cummings v Granger [1977] QB 397 and Curtis v Betts [1990] 1 WLR 459 and by the Court of Appeal in the instant case. The fact that an animal's behaviour, although not normal behaviour for animals of that species, was nevertheless normal behaviour for the species in the particular circumstances does not take the case outside section 2(2)(b).

    48. I also agree with the decision of the Court of Appeal on the facts in the present case. Horses are large and heavy animals. But it was not this innate physical characteristic of the defendants' horses which caused the road accident. The horses escaped because they were terrified. They were still not behaving ordinarily when they careered over the main road, crashing into vehicles rather than the other way about. Hale LJ concluded that it was precisely because they were behaving in this unusual way caused by their panic that the road accident took place: see [2002] 2 WLR 566, 571. That conclusion, on the evidence, seems to me irrefutable and to be fatal to the case of Dr and Mrs Henley. I would dismiss this appeal.

LORD SLYNN OF HADLEY

My Lords,

    49. Section (2) of the Animals Act 1971 provides as follows:

"(1)  Where any damage is caused by an animal which belongs to a dangerous species, any person who is a keeper of the animal is liable for the damage, except as otherwise provided by this Act.

(2)  Where damage is caused by an animal which does not belong to a dangerous species, a keeper of the animal is liable for the damage, except as otherwise provided by this Act, if—

(a) the damage is of a kind which the animal, unless restrained, was likely to cause or which, if caused by the animal, was likely to be severe; and

(b) the likelihood of the damage or of its being severe was due to characteristics of the animal which are not normally found in animals of the same species or are not normally so found except at particular times or in particular circumstances; and

(c) those characteristics were known to that keeper…"

    50. It is not surprising that different Courts in cases before the present one should have taken different views as to the meaning of section 2(2)(b) of the Act; nor that different views should emerge in the present case. The meaning of that part of the sub-section is not at all obvious or clear.

    51. It is however clear that 'the horse' is not a dangerous species of animal within the meaning of section 6(2) of the Act. To be such it has to satisfy two conditions. The first is that the animal is not commonly domesticated in this country; the second is that fully grown animals of the species "normally have such characteristics that they are likely, unless restrained, to cause severe damage or that any damage they may cause is likely to be severe". Whatever the position under the second condition it is plain that 'the horse' is commonly domesticated in the British Islands. Parliament has thus clearly excluded 'the horse' from sub-section (1) of section 2 of the Act as being an animal belonging to a dangerous species in respect of which the keeper of such an animal is liable for damage caused unless it is within the exceptions provided by section 5 of the Act, whether or not there is any fault on his part or knowledge by him of the likelihood of damage being caused.

    52. In sub-section (2) however Parliament has also clearly intended that under the statute (and apart from any possible liability for negligence at common law which has been negatived in the present case) the keeper of an animal not belonging to a dangerous species should, subject to the exceptions, in some circumstances be liable for damage caused by the animal. But that liability is subject to three conditions being satisfied. Conditions (a) and (c) are not in issue on this appeal, (a) being accepted as having been satisfied, in my view rightly, since it is likely that the damage if caused would be severe, and (c) the knowledge of the owners having been found to exist by the trial judge so that that condition is also satisfied. One therefore begins the consideration of (b) on the basis that in this case there was a likelihood of the damage if caused being severe. The question is thus whether such likelihood was due to characteristics of the animal (this particular horse) not normally being found in animals of the same species (domesticated horses) or not normally so found except at particular times or in particular circumstances.

    53. On this point I have had the advantage of reading in draft the opinion of my noble and learned friend Lord Scott of Foscote. I gratefully refer to his statement of the facts of the case and to his citation from the relevant authorities other than the passages which I have set out in this opinion. In the result I can express my views more briefly.

    54. Put simply the question whether the damage caused is due to characteristics normally found or not normally found in the species involves an inquiry as to whether the behaviour causing the damage is normal or abnormal for the species of animal. There is really no problem with the first part of section 2(2)(b)—do animals normally (are they prone to) bite or kick? The problem is with the second part—does one cancel the double negative 'not normally…except' and ask whether what was done in the special circumstances was normal behaviour for the species as a general rule; or is the right approach to ask whether what was done was normal for the species in the particular circumstances even if it will be abnormal in the absence of such circumstances. In Cummings v Granger [1977] QB 397, the first of these approaches was adopted. It was followed in Curtis v Betts [1990] 1 WLR 459 and in the Court of Appeal in the present case and commends itself to the majority of your Lordships. I refer to two passages. In Cummings v Granger Lord Denning MR at page 404 G to H said:

"Those characteristics—barking and running around to guard its territory—are not normally found in Alsatian dogs except in circumstances where they are used as guard dogs. Those circumstances are 'particular circumstances' within section 2(2)(b). It was due to those circumstances that the damage was likely to be severe if an intruder did enter on its territory. Section 2(2)(c): those characteristics were known to the defendant. It follows that the defendant is strictly liable unless he can bring himself within one of the exceptions in section 5."

    In Curtis v Betts Slade LJ said as follows at page 464:

"The broad purpose of requirement (b), as I read it, is to ensure that, even in a case falling within requirement (a), the defendant, subject to one exception, will still escape liability if, on the particular facts, the likelihood of damage was attributable to potentially dangerous characteristics of the animal which are normally found in animals of the same species. The one exception is this. The mere fact that a particular animal shared its potentially dangerous characteristics with other animals of the same species will not preclude the satisfaction of requirement (b) if on the particular facts the likelihood of damage was attributable to characteristics normally found in animals of the same species at times or in circumstances corresponding with those in which the damage actually occurred".

    On the other hand in Breeden v Lampard [21 March 1985 unreported] approved by Pill LJ in Gloster v Chief Constable of Greater Manchester Police [24 March 2000 unreported] the second approach was adopted. In Breeden v Lampard Lloyd LJ said at pp 9-10:

"The essential condition for liability now is that the characteristic which is known to the owner must be a characteristic which is abnormal for the species . . . If liability is based on the possession of some abnormal characteristic known to the owner, then I cannot see any sense in imposing liability when the animal is behaving in a perfectly normal way for all animals of that species in those circumstances, even though it would not be normal for those animals to behave in that way in other circumstances, for example, a bitch with pups or a horse kicking out when approached too suddenly, or too closely, from behind.

"In my view the purpose of the concluding words of s. 2(2)(b) of the Act may therefore be rather different. They may be designed to meet an argument by an owner:

"My horse did not have any abnormal characteristics even though it was liable to kick out all the time, because all horses are liable to kick out some of the time e.g. when crowded from behind"—in other words, the concluding words are refining what is meant by abnormality, not imposing a head of liability contrary to the main thrust of s. 2(2)(b) of the Act."

    55. In Breeden Oliver LJ said at p.12:

"for my part I share with my Lord, Lord Justice Lloyd, a puzzlement as to what is the meaning of the section because, like him, I cannot believe that Parliament intended to impose liability for what is essentially normal behaviour in all animals of that species".

    56. In Gloster v Chief Constable of Greater Manchester Police Pill LJ said:

"I respectfully agree with Lloyd LJ that the section is not concerned with animals behaving in a perfectly normal way for animals of the species or sub-species."

    57. Looking at the words in the context of the Act but without reference to the extracts from Hansard and the Report of the Law Commission to which your Lordships have been referred, it seems to me that the purpose of the legislation is, subject to specific exceptions, to impose strict liability for dangerous animals but to distinguish between normal and abnormal behaviour for non-dangerous animals. What is normal has to be considered (i) as a general rule and (ii) as an event in particular circumstances. The object of the provisions as I see it is to exclude strict liability not only for behaviour which is normal in the normal circumstances but also behaviour which is normal in particular (i.e. abnormal) circumstances, even if such behaviour would be abnormal in normal circumstances. There is thus strict liability for abnormal behaviour in normal circumstances and also for behaviour which is abnormal in abnormal (or particular) circumstances subject of course to para.(a) and (b) being satisfied. I do not think that the words used are intended to convert what is normal in abnormal circumstances to being abnormal in those circumstances because it would be abnormal in normal circumstances.

    58. If Parliament had intended that the keeper of a non-dangerous animal should be liable where the damage is caused by characteristics or behaviour which are normal at particular times, but not generally, (so that in effect the animal is to be treated as dangerous) it would have been sufficient to state quite simply that there should be liability if the likelihood of there being damage or if its being severe was due to characteristics of the animal which are not normally found in animals of the same species. The addition of the second part of (b) "are not normally so found except at particular times or in particular circumstances" should be read as meaning that if the characteristics are normally found in animals of the same species at particular times or in particular circumstances then there is no liability. It is only if the characteristics are abnormal at such times that liability attaches subject to paras. (a) and (c) being satisfied.

    59. To summarise, the intendment of para. (b) is that if the animal does what is normal for the species (a) usually or (b) only in special circumstances or at special times then it should not be treated as dangerous and there should be no strict liability, it being always remembered that liability in negligence is preserved.

    60. The Report of the Law Commission "Civil Liability for Animals" (1967) clearly indicated that the view of the Commission was that there should be strict liability for damage caused by an act which was not normal to the species in ordinary circumstances even if in certain circumstances it would be normal for members of the species so to act. Its draft Bill therefore accepted liability if "the animal has such characteristics that it is likely, unless restrained, to cause damage of that kind or that any damage of that kind that it may cause is likely to be severe" and if the keeper knew of those characteristics. There was no provision reflecting or setting out the section 2(2)(b) which found itself in the Act. This and the Law Commission's statements in paras. 18, 19 and 91(iv) thus support the respondent's case. The draft Bill was debated in 1969 but after the election, resulting in a change of government, section 2(2)(b) was added. A change from the Law Commission's proposal was apparently intended and it is not to be assumed that Parliament in the second Bill before the new Parliament intended to follow fully the Commission's proposal. Moreover I do not find the debates in Parliament to which your Lordships have been referred helpful as to what is the true import of section 2(2)(b). I would accordingly allow the appeal and I accept the application of the law as he and I see it to the facts which is proposed by my noble and learned friend Lord Scott of Foscote.

    LORD HOBHOUSE OF WOODBOROUGH

    My Lords,

    61. In agreement with my noble and learned friends Lord Nicholls of Birkenhead and Lord Walker of Gestingthorpe, I would dismiss this appeal. I am grateful to them for their very full discussion of the previous decisions of the courts and the arguments which have been advanced on this appeal which I will not repeat and will more briefly explain my reasons for upholding the decision of the Court of Appeal in this case.

    62. The salient facts of this case are not now in dispute. Horses when severely frightened are liable mindlessly to panic even to the extent of self destruction. It is normal for horses when sufficiently alarmed by a threat to attempt to flee from that threat, to ignore obstacles in their path, and when once able to flee to continue in flight for a considerable distance even beyond the point at which the perceived threat was detectable. It was also normal that, when encountering on the main road a further aversive situation beyond that which had caused their escape, the horse's fear would not only continue but heighten. When stressed, nervous activity increases and physiological changes occur. (See Agreed Facts, paragraph 3.6.) The defendants, the owners and statutory keepers of the three horses, knew that their horses would display these characteristics if so seriously frightened as to panic.

    63. On the night in question, at between 12.0 midnight and 12.30am, the horses were seriously frightened by some unascertained cause. They panicked. They charged down and flattened the fence of their field. They fled towards the A380 main road where they encountered motor traffic. This aggravated their existing state of panic. In this state of panic, two of the horses crashed into cars. In one incident, there were no personal injuries but the car was a write-off and the horse was killed. In the other, the horse charged into the side of the car of Mr Mirvahedy, the plaintiff in this case. The roof of the car was peeled off. Mr Mirvahedy suffered serious head and facial injuries. This horse was also killed by the impact. It is not now in dispute that the damage and injuries to Mr Mirvahedy and his car were caused by the horse and were severe.

    64. The question whether the defendants are liable to the plaintiff is to be answered by construing and applying the Animals Act 1971. It is not now suggested that the defendants acted unreasonably or were negligent. Nor is it suggested that the plaintiff was in any way to blame.

    65. In my view the relevant provisions of the Act have a plain and clear meaning. I do not see any justification for having regard to parliamentary materials nor do I consider that the materials to which your lordships were referred are sufficiently clear in themselves to be acceptable as aids to the construction of the statute. At least one parliamentary statement was patently based upon a misunderstanding.

    66. The Act was a reforming act and followed from Report No.13 of the Law Commission: Civil Liability for Animals (1967). Without adopting all the recommendations of the Law Commission, the Act completely recasts the previous law but has retained a recognisable structure derived from the previous law. Thus it retains a distinct category for wild animals "not commonly domesticated in the British Islands", s.6(2)(a) and s.2(1); and has a residual category which makes use of the former scienter rule based on the keeper's knowledge of the particular animal's actual characteristics, s.2(2)(c).

    67. Another feature of the Act is that it uses a double-barrelled concept of dangerousness with alternative criteria either of which suffices. The first is the familiar characteristic that the animal or its species is, unless restrained, likely to cause severe damage; this corresponds to what has sometimes been called a vicious propensity. The second is directed not to the animal's propensities, be they vicious or benign, but to the consequences of anything it may do. Thus the alternative criterion is that it is an animal of which it can be said that "any damage [it] may cause is likely to be severe". These two alternative criteria are used in conjunction with the criterion of non-domestication to define what is a dangerous species of animal in s.6(2). Using the first alternative, a tiger is a dangerous animal. It is likely, unless restrained, to cause severe injuries to humans: that is its nature. Using the second alternative, an Indian elephant is a dangerous animal, not because it is likely to injure any one, but because, if it does, the injury is likely, as a result of its weight and bulk, to be severe: cf Behrens v Bertram Mills Circus [1957] 2 QB 1. This is a statement about its physical capacity to injure and its inability to limit the consequences of that capacity not about its inclination to injure. In s.2(1) there is a strict liability for damage caused by dangerous animals as defined in s.6(2). In s.2(2) there is a scienter liability for any damage caused by any other animal which is, inter alia, damage of a kind which the animal in question was, unless restrained, likely to cause or which, if caused by that animal, was likely to be severe: s.2(2)(a).

    68. This is the starting point for the legal question which has arisen in this case. The damage to Mr Mirvahedy and his car by the panicking horse when it charged into his car and landed on its roof was and was likely to be severe, (2)(a). Similarly the keepers of the horse knew of the characteristics of horses in general and their horse in particular which made such damage a likely consequence of such conduct in a state of panic, (2)(c). It is accepted that it is not a normal characteristic of horses to cause such damage. They may have the capacity to kill a man by kicking him on the head but it is not likely that any normal conduct of theirs will lead to that result nor that they have a normal propensity to attack human beings. If it had been the case that the horse in question was known to have characteristics which made such injuries likely in the ordinary course, there would be no question but that the requirements for liability under s.2(2) would have been satisfied and the defendants would be liable. But that is not this case. The question is whether the other alternative in s.2(2)(b) is satisfied: whether the likelihood of the damage or of its being severe was due to characteristics of the horse which are not normally found in horses except at particular times or in particular circumstances.

    69. Horses are not normally in a mindless state of panic nor do they normally ignore obstacles in their path. These characteristics are normally only found in horses in circumstances where they have been very seriously frightened. It is only in such circumstances that it becomes likely that, due to these characteristics, the horse will cause severe damage. This case clearly comes within the words of s.2(2)(b). There is no ambiguity either about the facts of this case or about the meaning of paragraph (b).

    70. The Report of the Law Commission supported such a conclusion in its recommendations for the retention of the scienter principle: see paragraphs 17, 18 and 91 of the Report. Using the example of a bitch with puppies, the Commission said (at paragraph 18(ii)):

"In our view the fact that a particular animal belonging to a non-dangerous species shares [dangerous] characteristics with other animals within the species, either at a particular age, at certain times of the year or in special conditions, should not preclude liability where the keeper knows of the presence of these characteristics in the animal at the time of the injury. If the keeper of a bitch with a litter knows that it is prone to bite strangers, then even if this is a common characteristic of bitches at such a time, we think that the keeper should be strictly liable, subject to the permissible defences ....."

    71. The contrary argument seems to be based upon the view that any normal behaviour of a domesticated animal should not give rise to liability. This point was clearly put in the judgment of Lloyd LJ in Breeden v Lampard from which my noble and learned friend Lord Nicholls has already quoted. It is true that there is an implicit assumption of fact in s.2(2) that domesticated animals are not normally dangerous. But the purpose of paragraph (b) is to make provision for those that are. It deals with two specific categories where that assumption of fact is falsified. The first is that of an animal which is possessed of a characteristic, not normally found in animals of the same species, which makes it dangerous. The second is an animal which, although belonging to a species which does not normally have dangerous characteristics, nevertheless has dangerous characteristics at particular times or in particular circumstances. The essence of these provisions is the falsification of the assumption, in the first because of the departure of the individual from the norm for its species, in the second because of the introduction of special factors. Criticisms can be, and have been, made of the drafting of paragraphs (a) and (b) of s.2(2); but they should not be made, and are not justified, in this respect of the drafting of paragraph (b). It does not lack coherence.

    72. The statute, in this respect following the recommendation of the Law Commission, had to reflect a choice as to the division of risk between the keeper of an animal and members of the general public. Neither is blameworthy but it is the member of the public who suffers the injury or damage and it is the keeper who knows of the characteristics of the animal which make it dangerous and liable to cause such injury or damage. The element of knowledge makes the choice a coherent one but it, in any event, was a choice which it was for the Legislature to make.

    73. For these reasons, which accord in most respects with those given by my noble and learned friend Lord Nicholls and to be given by my noble and learned friend Lord Walker, I would dismiss the appeal.

LORD SCOTT OF FOSCOTE

My Lords,

    74. The issue in this appeal is the extent of the strict liability for damage imposed by the Animals Act 1971 on the owners, or keepers, of animals which do not belong to a dangerous species. The issue turns on the correct construction of section 2(2)(b), and to a lesser extent, section 2(2)(a), of the Act. Over the 30 years or so that the Act has been in force different judicial views have been expressed on this issue of construction. However this is the first time that the issue has reached your Lordships' House. An authoritative ruling is overdue.

    75. The essential facts of the case can be shortly stated. The appellants, Dr and Mrs Henley, lived at Haldon Lodge, Chudleigh in Devon. They owned a horse, a 5 year old of 15.2 hands, and two ponies one 14.1 hands, the other 11.2 hands. They kept the three animals in a six acre field adjacent to their house. The field was enclosed by a post and barbed wire fence. In 1994 an electric fence, supported mainly on outriggers attached to the existing fence posts but in places supported also by plastic posts, had been added.

    76. Shortly after midnight on Thursday 29 August 1996 the three horses escaped from their field at its north western corner. The reason for their escape and the circumstances that led to their escape remain unknown but the scene, as it appeared the next morning, was described by the trial judge, Judge O'Malley, as follows:

"the fence at the northern corner of the field was found to have been flattened outwards. The corner post and the barbed wire were flat on the ground and the electric fence had been pulled through as the horses exited. The electrotape and the electrorope had snapped. Thick vegetation was trampled in a diagonal line from the corner of the field to the lane. Within the field some plastic fence posts had been uprooted and there was an unusual and extensive area of trampling right into the apex of the corner with fresh hoofprints." (p. 3 of his judgment).

    77. The judge, having heard the evidence of experts, concluded that the horses had been frightened by something or someone in their field and, in their fright, had bolted into and through the fence. They were normally docile horses and nothing in the previous behaviour of any of them had indicated a propensity to try and escape from their field.

    78. The horses' field is bounded on its western side by a rough lane. Some 300 yards on from the field the lane runs into a minor local road which, in turn, after a mile or so runs into the A380. The A380 is a major dual carriageway road.

    79. The respondent, Mr Mirvahedy, was driving from Torquay to Exeter on the A380 when, at about 12.30 am on the Thursday night that the horses escaped, his car came into collision with the 15.2 hand horse. The horse was killed. Mr Mirvahedy's car was severely damaged and he, himself, suffered serious injuries. The 14.1 hand animal came into collision with another car on the A380. It, too, was killed. The 11.2 hand pony was caught by some third party and placed in an adjacent field from whence he was retrieved by his owners the next morning. The evidence was that when retrieved the pony was still very distressed and nervous.

    80. Mr Mirvahedy commenced an action for damages against Dr and Mrs Henley. He claimed both in negligence and under the Animals Act 1971. His negligence claim failed. Judge O'Malley concluded that "these normally docile horses were adequately contained by the fence in question" and that "the fencing at the point of the escape did not fall below the standard required of a reasonably careful and prudent owner of horses such as those with which we are concerned" (pp. 8 and 9 of the judgment). There was no appeal against these findings.

    81. Accordingly, Mr Mirvahedy's ability to recover from Dr and Mrs Henley damages for the injuries he had suffered and for the damage to his car depends on his ability to impose on the Henleys the strict liability for damage prescribed by the 1971 Act. My Lords, in considering whether or not Mr Mirvahedy is able to do so, it is important to bear in mind first that the Henleys have been exonerated from negligence and, second, that the horse that caused the damage was a normal docile animal with no mischievous propensity, or, at least, none of any relevance to the accident with which this case is concerned.

    82. It is not in dispute that the purpose of the Act was to amend the common law relating to civil liability for damage done by animals. The Act addressed (inter alia) three rules of the common law; first, the rule imposing strict liability on the keeper of an animal ferae naturae, ie. wild, for damage caused by the animal; second, the rule imposing strict liability on the keeper of an animal mansuetae naturae, ie. tame, for damage caused by the animal attributable to a vicious, mischievous or fierce propensity which the keeper knew the animal possessed (the so-called scienter rule); and, third, the rule barring an action in negligence against the owner of livestock which stray on to the highway and there cause an accident. The third rule was addressed and reformed by the Act, thus enabling Mr Mirvahedy's negligence claim to proceed to trial, but is of no further relevance to the present case.

    83. Section 1 of the Act (inter alia) replaced the first and second above-mentioned common law rules by section 2 of the Act. Section 2 has the headline "Liability for damage done by dangerous animals". I draw attention to the adjective "dangerous". The section provides as follows:

"(1)  Where any damage is caused by an animal which belongs to a dangerous species, any person who is a keeper of the animal is liable for the damage, except as otherwise provided by this Act.

Where damage is caused by an animal which does not belong to a dangerous species, a keeper of the animal is liable for the damage, except as otherwise provided by this Act, if—

(a) the damage is of a kind which the animal, unless restrained, was likely to cause or which, if caused by the animal, was likely to be severe; and

(b) the likelihood of the damage or of its being severe was due to characteristics of the animal which are not normally found in animals of the same species or are not normally so found except at particular times or in particular circumstances; and

(c) those characteristics were known to that keeper or were at any time known to a person who had charge of the animals as that keeper's servant or, where that keeper is the head of a household, were known to another keeper of the animal who is a member of that household and under the age of sixteen."

    84. Section 5(1), (2) and (3) of the Act provides certain exceptions from liability under section 2:

"(1)  A person is not liable under sections 2 to 4 of this Act for any damage which is due wholly to the fault of the person suffering it.

A person is not liable under section 2 of this Act for any damage suffered by a person who has voluntarily accepted the risk thereof.

A person is not liable under section 2 of this Act for any damage caused by an animal kept on any premises or structure to a person trespassing there, if it is proved either—

(a) that the animal was not kept there for the protection of persons or property; or

(b) (if the animal was kept there for the protection of persons or property) that keeping it there for that purpose was not unreasonable."

    85. None of these exceptions applies in the present case but their contents may assist in understanding the extent and nature of the strict liability intended to be imposed under section 2.

    86. The expression "dangerous species" is defined by section 6(2).

"(2)  A dangerous species is a species

(a) which is not commonly domesticated in the British Islands; and

(b) whose fully grown animals normally have such characteristics that they are likely, unless restrained, to cause severe damage or that any damage they may cause is likely to be severe."

Under section 6(3), a person is a keeper of an animal if he is the owner of it. And, in section 11, "species" is defined as including "sub-species and variety".

    87. The verb "domesticate" has a number of possible meanings. One of these, according to the Shorter Oxford English Dictionary, 3rd Ed., and that which I think best accords with the context of section 6, is "to tame or bring under control". So species of animals which are commonly tamed and brought under control in this country cannot belong to a "dangerous species". Animals which would, under the old law, have been mansuetae naturae in this country are excluded. But an animal not commonly domesticated (in this country), for example, an animal ferae naturae under the old law, does not necessarily belong to a "dangerous species". The animal must belong to a species which satisfies paragraph (b) of section 6(2). Thus, for example, a deer, kept in captivity, would not, in my opinion, satisfy the section 6(2)(b) requirement and would not belong to a "dangerous species" for section 2 purposes.

    88. The language of section 2(2) has been the subject of judicial excoriation in a number of cases. These have been referred to by Hale LJ in paragraph 18 of the judgment under appeal in the present case and it is not necessary to repeat them. It is not in dispute that the language of paragraph (b) of section 2(2) is ambiguous. The ambiguity results from uncertainty about the intended function of the concluding words of paragraph (b): "or are not normally so found except at particular times or in particular circumstances".

    89. The words may be read as providing two alternative ways in which the requirements of the paragraph can be satisfied. So read, the paragraph requires the likelihood of the damage or of its being severe to be due either (i) to characteristics of the animal which are not normally found in animals of the same species, or, alternatively (ii) to characteristics of the animal which are normally found in animals of the same species at particular times or in particular circumstances.

    90. There are two features of this construction that I think need to be highlighted. First, the alternative requirement, as expressed in (ii) (the second limb requirement), involves cancelling out the double negative to be found in the statutory language. "…. not normally so found except at particular times …." etc. becomes "normally so found at particular times …." etc. The statutory injunction is to look for characteristics not normally found. There is no expressed injunction to look for characteristics which are normally found. But the second limb requirement imposes that injunction.

    91. Second, if the second limb requirement is applied to an actual case in which an animal has caused damage, this construction requires an inquiry as to whether at the particular time or in the particular circumstances that the animal caused the damage it would have been normal for any animal of the same species to cause the same or similar damage. In other words the statutory reference to "at particular times or in particular circumstances" is treated as a reference to the particular time at which or the particular circumstances in which the actual animal caused the damage in question.

    92. On the other hand, paragraph (b) may be read as imposing a single, composite requirement, namely that the likelihood of the damage or of its being severe was due to characteristics of the animal which are not normally found in animals of the same species except at particular times or in particular circumstances. On this reading the function of the concluding words of paragraph (b), "or are not normally so found except at particular times or in particular circumstances", is not to impose a second test but to exclude a defence based on the undoubtedly true proposition that all dogs bite sometimes, all horses kick sometimes, all bulls charge sometimes, etc. On this reading the reference to "at particular times or in particular circumstances" is a reference to times or circumstances which necessarily must be different from those in which the animal which has caused the damage has done so. If, in the circumstances in which the animal had caused the damage, it would not have been normal for an animal of its species to do so, there will be strict liability (subject to the paragraphs (a) and (c) requirements) even though at other times and in other circumstances it would be normal for an animal of its species to cause similar damage.

    93. The issue of construction of paragraph (b) depends, in my opinion, on identifying the function intended by Parliament to be served by the concluding words of the paragraph.

    94. The paragraph (b) issue is not the only issue of construction arising under section 2(2). Paragraph (a), too, raises issues of construction prompted by the facts of this case. It is convenient to refer to these before returning to the paragraph (b) issue.

    95. Section 2(2)(a) requires the damage either to be "of a kind which the animal, unless restrained, was likely to cause", or to be "of a kind … which, if caused by the animal, was likely to be severe." This language is virtually identical to that in section 6(2)(b). The similarity is plainly not coincidental. A wild animal, kept in captivity, does not belong to a "dangerous species" and cannot attract the strict liability imposed by section 2(1) unless it is likely, if free of restraint, to cause severe damage or unless any damage it causes is likely to be severe— unless, in short, it is a dangerous animal. The word "likely" in the content of section 6(2)(b) should, in my opinion, be given its natural meaning of "to be reasonably expected" (see The Concise Oxford Dictionary, 9th Ed., p. 789). It would not, in my opinion, be enough that the animal, unless restrained, could or might cause severe damage. Any animal could do that. You could trip over an escaped dormouse and break your neck falling downstairs. Nor would it be enough that damage, if caused by the animal, could or might be severe. The statutory language requires any damage caused by the animal to be likely to be severe. A mere possibility would not suffice. If a reasonable expectation, whether that the animal would, if free of restraint, cause severe damage or that any damage caused by the animal would be severe, were absent, the requirements of paragraph (b) of section 6(2) would not be satisfied. The animal would not belong to a "dangerous species". In short, it would not be a dangerous animal. So strict liability under section 2(1) would not be attracted.

    96. The word 'likely' should, in my opinion, be accorded the same meaning in section 2(2)(a) as it has in section 6(2)(b). If a commonly domesticated animal causes damage, the first two questions, if strict liability is to be imposed, are whether the damage is of a kind that the animal, unless restrained, was likely to cause or whether the damage is of a kind which, if caused by the animal, was likely to be severe. The answer to these questions cannot be answered by simply referring to the seriousness of the damage actually caused in the case in question. To do so would be to ignore the inclusion in the statutory language of the word "likely". If a large domesticated animal, say a horse or a bullock, finds itself loose and unrestrained in a public place it may cause personal injury or injury to property. But is it likely to do so? If it does cause personal injury or injury to property is injury of that kind likely to be severe? Neither of these questions can be answered simply by saying that the animal has in fact caused severe personal injury to the complainant or has in fact caused severe damage to his property. If that were the right approach paragraph (a) could simply have read "or which, if caused by the animal, was severe".

    97. In Smith v Ainger, an unreported case in the Court of Appeal in which judgment was given on 16 May 1990, a large delinquent dog attacked the plaintiff's dog and, in the process, knocked over the plaintiff causing him to break a leg. The only judgment in the case was given by Neill LJ with whom the other two members of the court expressed agreement. Neill LJ directed himself first to the meaning to be given to the word "likely" in section 2(2)(a). He rejected "probable" or "more probable than not" as correct and preferred "such as might happen" or "such as might well happen". I would respectfully agree with the Lord Justice's rejection of "probable" and "more probable than not" but am unable to agree that "such as might happen", a phrase consistent with no more than a possibility, can be right. A mere possibility is not, in my opinion, enough. I have suggested "reasonably to be expected" as conveying the requisite meaning of "likely" in paragraph (a). But it may be that there is no material difference between "reasonably to be expected" and Neill LJ's "such as might well happen".

    98. In my opinion, there has been insufficient attention paid in the present case to the requirements of paragraph (a). It seems to have been assumed that because Mr Mirvahedy suffered serious personal injuries caused by the escaped horse and that considerable damage was caused to his car, the requirements of paragraph (a) were shown to be satisfied. My own impression, however, is that a horse loose on the highway does not usually result in damage to third parties, that if damage to third parties does result the damage is not usually severe, no more, perhaps, than a dent to a car, and that the cases in which serious injury or damage results are fortunately few and far between.

    Jaundrill v Gillett, also unreported, heard by the Court of Appeal on 16 January 1996, was, like the present case, a case in which horses had found their way from their field to a highway. In Jaundrill the reason for this was that some malicious person, having opened the gate, had driven the horses out of their field. As in the present case, there had been no negligence on the part of the horses' keeper. The plaintiff's car collided with two of the horses. The plaintiff suffered some personal injury and his car was damaged. He sought compensation pursuant to the strict liability imposed by section 2(2) of the 1971 Act. The defendant conceded that, having regard to the weight of each horse, damage caused by the animal was likely to be severe and that paragraph (a) was satisfied (see Russell LJ's judgment at p. 3 of the transcript). In the present case, Judge O'Malley said that

"… even if the likelihood of the horses causing personal injury, unless restrained, was remote, there was a likelihood that such injury, if caused, would be severe. As was stated by Russell LJ in Jaundrill v Gillett, which also concerned escaping horses, this criterion gave rise to no difficulty having regard to the weight of the animal."

    100. In the Court of Appeal there appears to have been no challenge to Judge O'Malley's conclusion that the "likelihood" requirement of paragraph (a) was satisfied. And, indeed, no challenge on that point has been raised before your Lordships. In deciding this appeal, therefore, your Lordships have no alternative but to proceed on the footing that the judge's conclusion on this point was correct. But I would wish to express my reservations about it.

    101. The meaning to be attributed to the adverb "likely" is not the only point of construction arising under paragraph (a). It is not, in my opinion, entirely clear what is meant in paragraph (a) by "damage …. of a kind ….". In a case like the present, where personal injury has been caused by a collision with a bolting horse, does it mean personal injury of a kind likely to be caused by collision with a bolting horse? Or does it merely mean personal injury of a kind likely to be caused by a horse unless restrained? In the former case the damage is obviously likely to be severe; in the latter case the likelihood of damage or of damage being severe is not apparent and might at least warrant some evidence. This point, too, has not been addressed in the present case and cannot, therefore, be taken any further on this appeal.

    102. Your Lordships have been given a Pepper v Hart invitation and taken to passages in Hansard recording comments about the Bill made in 1969 and 1970 during its progress through Parliament. My Lords, the passages in question are, in my opinion, inconclusive and do not provide any clear answer to the question as to the intended function of the concluding words of paragraph (b). That being so, the Hansard passages should be set aside and the statutory words in question given a function consistent with the language used, with the general scheme of the Act and with the reasonable presumption that Parliament does not intend absurd results.

    103. The general scheme of the Act was to draw a distinction, so far as the imposition of strict liability is concerned, between dangerous animals and other animals: see, for instance, the definition of "dangerous species" in section 6. An animal may be dangerous either because it belongs to a dangerous species or, if it does not belong to a dangerous species, because it has particular dangerous propensities of its own. In relation to an animal which belongs to a dangerous species, the Act imposes strict liability for damage it causes so as, in effect, to require the keeper, subject to the section 5 exceptions, to be the insurer that the animal will not cause damage to third parties. In relation to an animal which does not belong to a dangerous species, the strict liability imposed by the Act is severely limited. The requirements of paragraphs (a), (b) and (c) of section 2(2) must each, cumulatively, be satisfied before the keeper will come under strict liability for damage caused by the animal. It is a reasonable inference that the Parliamentary intention in relation to these animals was that the keeper's primary liability for damage would be a liability in negligence. Strict liability would lie only in special circumstances.

    104. It would have been very easy for the Act to have imposed strict liability on the keeper of an animal which did not belong to a dangerous species for any damage caused by any dangerous propensity of the animal, with the paragraph (c) requirement of knowledge constituting the only limit on that strict liability. The Act did not do so. The opening words of paragraph (b) impose strict liability only in respect of damage attributable to a particular dangerous propensity of the animal, abnormal for the species to which the animal belongs.

    105. It is, however, plainly possible for an animal not belonging to a dangerous species to possess, in common with all other animals of its species, dangerous propensities at particular times or in particular circumstances. These propensities need not abnormal. An example often given is that of a bitch with pups. It is not a normal characteristic of a dog to bite a stranger who approaches it. But for a bitch to bite a stranger who approached and tried to handle her pups would be normal behaviour. The example does not perhaps matter for it is clear enough that there are particular circumstances in which a normal, docile, domesticated animal is likely to display dangerous characteristics. The issue is whether Parliament, in enacting paragraph (b) intended the keeper to attract strict liability for damage done by the animal in those circumstances.

    106. The dual test construction of section 2(2)(b), described in paragraph 16 above, was accepted by both Judge O'Malley and the Court of Appeal in the present case. The construction is supported by the Court of Appeal decisions in Cummings v Granger [1977] QB 397 and Curtis v Betts [1990] 1 WLR 459.

    Cummings v Granger concerned a young, untrained Alsatian dog. The dog was turned loose at night in a breaker's yard to deter intruders. The plaintiff entered the yard and was bitten by the dog. The trial judge held that paragraphs (a), (b) and (c) of section 2(2) were all satisfied and found in favour of the plaintiff. But the Court of Appeal held that the owner of the dog had a good defence under section 5(3) and allowed the appeal. As to section 2(2)(b), all three members of the Court appear to have accepted the dual test construction of paragraph (b) (see Lord Denning MR at 404G, Ormrod LJ at 407B and Bridge LJ at 409D/E). But none found the meaning of the paragraph to be clear: Lord Denning said that it would "give rise to several difficulties in future" and Ormrod LJ described its language as "remarkably opaque". These were ex tempore judgments and the alternative, single test, construction, restricting strict liability to liability for damage caused by the animal displaying characteristics not normally found in animals of its species, was not put to the Court.

    108. In my opinion, the judge and the Court of Appeal in Cummings v Granger were right, whichever be the right construction of paragraph (b), to find that the requirements of the paragraph were satisfied. It is not a normal characteristic of an Alsatian, or of any species of domestic dog that I have ever heard of, to make an unprovoked attack of the sort made by the Cummings v Granger Alsatian. There was much talk in the case of the dog being young and untrained and kept as a guard dog and of Alsatians having territorial instincts which make them likely to attack strangers who venture into their territory. The norm, however, for the purposes of application of paragraph (b), construed according to the single test construction, would not be a young, untrained dog. It would be an adult dog which had received the degree of training usual for the species. A puppy is not a "sub-species or variety" of dog (see s. 11). Nor is a guard dog a "sub-species or variety". The decision in Cummings v Granger did not turn on which of the two alternative approaches to construction of paragraph (b) was correct and the Court of Appeal did not address that issue.

    Curtis v Betts was another dog case. A bull mastiff, in course of being taken from his owner's house to his owner's car, was approached by a 10 year old child and attacked and injured the child. Liability under section 2(2) of the Act was found by the trial judge to be established. The defendant's appeal was dismissed. The Court of Appeal adopted the dual test construction of paragraph (b) but the result would, in my opinion, have been the same whichever construction had been adopted. Slade LJ analysed paragraph (b) in these terms:

"The broad purpose of requirement (b), as I read it, is to ensure that, even in a case falling within requirement (a), the defendant, subject to one exception, will still escape liability if, on the particular facts, the likelihood of damage was attributable to potentially dangerous characteristics of the animal which are normally found in animals of the same species. The one exception is this. The mere fact that a particular animal shared its potentially dangerous characteristics with other animals of the same species will not preclude the satisfaction of requirement (b) if on the particular facts the likelihood of damage was attributable to characteristics normally found in animals of the same species at times or in circumstances corresponding with those in which the damage actually occurred." (p. 464).

    110. Slade LJ's explanation of section 2(2)(b) provides strong support for the Court of Appeal's approach to paragraph (b) in the present case. But, as with Cumming v Granger, I doubt whether the alternative, single test approach to paragraph (b), would have led to any different result. It is not quite clear from the report of the case what the evidence before the court was as to the characteristics of bull mastiffs, but the proposition that it is a normal characteristic of bull mastiffs to make unprovoked attacks, such as was made on the child in Curtis v Betts, I find an astonishing one. The case did not, in my opinion, need to come within Slade LJ's explanation in order to satisfy paragraph (b).

    111. Stuart-Smith LJ, too, addressed himself to paragraph (b) on the footing that it contained two limbs. The first limb, he said, dealt with "permanent characteristics", the second with "temporary characteristics". I respectfully doubt whether this is a helpful distinction. What are "permanent" characteristics? There are no behavioural characteristics of animals that are on display all the time. Every behavioural characteristic is displayed when appropriate circumstances arise. All behavioural characteristics possessed by any animal are temporary, arising when the circumstances are apt and dormant when they are not. But I have no doubt whatever but that in Curtis v Betts the Court of Appeal (and the trial judge) reached the right decision.

    112. The explanation of the concluding words of paragraph (b) given by Slade LJ in Curtis v Betts, and accepted and applied by the Court of Appeal in the present case, seems to me to be inconsistent with the paragraph taken as a whole. I would respectfully agree that the broad purpose of the paragraph was to ensure that the keeper of the animal would escape strict liability if, on the particular facts, the likelihood of damage was attributable to potentially dangerous characteristics of the animal which would normally be found in animals of the same species (see Slade LJ, at p. 464). But the "one exception" to this, identified by Slade LJ and based on the concluding words of the paragraph, seems to me wholly to undermine and frustrate that broad purpose. If an animal causes damage it will necessarily do so at a particular time and in particular circumstances. Damage cannot be caused in the abstract. There must necessarily be a time and a context. It is not possible, in my opinion, to construct a case in which an animal acting normally has caused damage, a fortiori severe damage, otherwise than at a particular time or in particular circumstances. Slade LJ's exception to the "broad purpose" that he rightly, in my opinion, attributed to paragraph (b) would prevent the keeper of the animal that had caused the damage from ever escaping strict liability for damage attributable to normal characteristics of the species — always assuming that the requirements of paragraphs (a) and (c) were satisfied. I do not think this can correspond with Parliament's intention.

    113. The problem about Slade LJ's exception to which I have just referred can, arguably, be avoided by giving some restricted meaning to "at particular times or in particular circumstances". If "particular" is given a meaning of "unusual" or "special", or some comparable meaning, there would not necessarily be strict liability for damage caused by an animal when acting normally. Strict liability would only be attracted if the animal was acting normally in "unusual" or "special" circumstances, or at an "unusual" or "special" time. This suggested solution to the problem would, to my mind, simply replace it with a worse one. What possible criteria could be applied to determine whether the time or the circumstances were "particular". Would the criteria be statistical? Or would the criteria be subjective? And if so, from whose point of view? The animal's? In my opinion, since every behavioural characteristic is only displayed in circumstances that prompt its display, the only sensible reading of "particular times" and "particular circumstances" is times or circumstances that prompt the display of the characteristic in question. But this reading produces the problem to which I referred in the last foregoing paragraph.

    114. Consider also some of the strange results to which this construction of the concluding words of paragraph (b) would give rise. Strict liability would be imposed for any damage caused by an animal when responding to any external stimulus in a manner entirely normal for its species. Take the case of a normal, docile horse in a field through which a public footpath runs. A mischievous individual using the footpath shoots a projectile at the horse with a catapult and hits the horse. The horse bolts and knocks over some other entirely innocent user of the footpath. The owner of the horse is likely to be well aware that any horse if shot by a projectile from a catapult would be likely to bolt. On the dual test construction of paragraph (b) — as set out in paragraph 16 above — the owner would have strict liability for the damage to the injured third party.

    115. Or, as a further example, take the case of mounted police endeavouring to control a demonstration. Police horses are very well trained and in general do not kick at those in their vicinity. They are accustomed to crowds and loud noises. But suppose an individual in the crowd were to jab a horse's rump with some sharp instrument. I doubt whether there is any horse so well trained that it would not respond to provocation of that sort by kicking out. If the kick were to connect with the miscreant, he would be barred by section 5(1) from claiming under the Act. But if the kick were to connect instead with some other member of the crowd, would the injured person have a strict liability claim against the police? The horse, in kicking out, would have displayed a characteristic normally found in horses into whose rumps a sharp instrument has been jabbed. A contention that the keeper of the horse was not aware of this characteristic would be unreal. On the dual test construction of paragraph (b) there would be strict liability.

    116. Consider also how the dual test construction of paragraph (b) would work in the case of animals, such as deer, which are not commonly domesticated in this country but which are not likely to cause severe damage while at large. Deer can hardly be regarded as belonging to a "dangerous species" as defined in section 6(2). But although deer are not commonly domesticated in this country, there are deer farms and deer parks where deer are kept in captivity either to provide pleasure to the beholder or, more prosaically, to provide a source of venison. From time to time deer are to be seen on our highways. From time to time deer on highways come into collision with motor vehicles. It is easy to imagine a deer on a highway at night, frightened by the noise and headlight glare of the vehicles, seeking to escape but in its panic fleeing in the wrong direction and colliding with a vehicle thereby causing damage — behaving, that is to say, not unlike the horses in the present case and with a similar result.

    117. If the deer were a wild deer, no one would be liable for the damage. But if the deer had escaped from a deer farm or deer park, its owner would, if the dual test approach to paragraph (b) is right, be liable notwithstanding that the animal had behaved in a manner entirely normal for a deer in the circumstances in which it found itself. This would produce the paradoxical situation in which on the one hand deer are removed by section 6(2)(b) from being categorised as a "dangerous species" but on the other hand an individual deer may impose strict liability on its keeper under section 2(2)(b) for damage caused by behaviour entirely normal for the species.

    118. My Lords, I cannot believe that Parliament intended paragraph (b) to have the effect described. To impose strict liability on the keeper of an ordinary domesticated animal, or of a non-dangerous wild animal held in captivity, for damage done by the animal when responding normally, as any member of its species would respond, to some external stimulus seems to me inconsistent with the apparent intention of the Act to draw a distinction between dangerous and non-dangerous animals and inconsistent, in particular, with the apparent purpose of paragraph (b) to limit strict liability for non-dangerous animals to damage attributable to abnormal characteristics. If there was intended to be strict liability for damage caused by an animal behaving in a manner normal for its species, one would have expected the Act to categorise the species as a dangerous species. The fact that an animal belongs to a species that falls outside the statutory definition of a dangerous species is, in my opinion, an indication that behaviour by the animal in a manner normal for the species was not intended by Parliament to attract strict liability. If there was intended to be strict liability for damage caused by normal behaviour of non-dangerous animals one would have expected that simple proposition to be simply stated rather than left to be produced by the literary device of turning the double negative in the concluding words of paragraph (b) into a positive.

    119. A further paradox that seems to me worth noting is that, if the dual-test construction of paragraph (b) is right, a professional keeper of animals will have a more extensive strict liability than an ignorant amateur. Take the present case. Dr and Mrs Henley were experienced horse owners. They were well aware of the natural tendency of horses to flee when startled or frightened and, in the case of several horses together, to do so as a herd. They are herd animals. The amateur, buying his first horse, might plead ignorance of these characteristics and hope to prevent the requirements of paragraph (c) from being satisfied. The Henleys could not, and did not, do so.

    120. I find it quite impossible to understand what legislative policy could be served by allowing a keeper's ignorance of the normal characteristics of the animal in his charge to permit him to escape the strict liability imposed on a responsible keeper who had made himself aware of those characteristics. The requirements of paragraph (c) only make sense, in my opinion, if the characteristics that have caused the damage in question are not normal to the species but are peculiar to the individual animal.

    121. The alternative single test approach to paragraph (b) as set out in paragraph 19 above is supported by dicta in two unreported decisions of the Court of Appeal, Breeden v Lampard, in which judgment was given on 21 March 1985, and Gloster v Chief Constable of Greater Manchester Police, in which judgment was given on 24 March 2000.

    Breeden v Lampard was the case where a horse, at a meet of the Atherstone Hunt, kicked out breaking the leg of the rider of another horse. It appears that the horses and riders were progressing down a road. Per Sir George Waller, at p 2 of the transcript:

"when the appellant was close behind the respondent the respondent's horse, Raffles, shuffled to the left and then kicked out, causing the appellant to suffer the broken leg."

Raffles was wearing a red ribbon on his tail. Such an adornment is a traditional warning to others that the horse is prone to kick and to keep their distance. The trial judge, Macpherson J, found that there had been no negligence on the part of Raffles' rider and that strict liability under section 2(2) had not been established. The Court of Appeal dismissed the appeal.

    123. On the section 2(2)(b) point, Sir George Waller took the view that Raffles had no characteristics not normally found in animals of the same species. The question whether Raffles had characteristics arising at a particular time or in particular circumstances did not in his opinion arise (see p 5A-B of the transcript) for the trial judge had found that his rider was not aware of any such characteristics. Sir George Waller did not, therefore, need to deal with the question whether the concluding words of paragraph (b) imposed a separate and alternative test of liability, and he did not do so.

    124. Lloyd LJ, however, dealt explicitly with the point. He said this:

"In the old law there was much debate as to whether the owner of an animal, not being an animal ferae naturae, was liable for injury caused by a vice natural to the species of that animal ….

But all that now has been swept away by s. 2(2)(b) of the new Act. The essential condition of liability now is that the characteristic which is known to the owner must be a characteristic which is abnormal for the species …. If liability is based on the possession of some abnormal characteristic known to the owner, then I cannot see any sense in imposing liability when the animal is behaving in a perfectly normal way for all animals of that species in those circumstances, even if it would not be normal for those animals to behave in that way in other circumstances, for example, a bitch with pups or a horse kicking out when approached too suddenly, or too closely, from behind. In my view, the purpose of the concluding words of s. 2(2)(b) …. may be designed to meet an argument by an owner:

'My horse did not have any abnormal characteristics even though it was liable to kick out all the time, because all horses are liable to kick out some of the time eg. when crowded from behind'.

In other words, the concluding words are refining what is meant by abnormality, not imposing a head of liability contrary to the main thrust of s. 2(2)(b) of the Act." (pp. 9 and 10 of the transcript).

In the final sentence of this citation, Lloyd LJ is attributing to the concluding words the same function as that suggested in paragraph 19 above.

    125. Oliver LJ expressed agreement with both his colleagues. He paraphrased paragraph (b) in terms virtually identical to those in paragraph 19 above (see p 11B of the transcript). He commented (p. 12 of the transcript):

"I cannot believe that Parliament intended to impose liability for what is essentially normal behaviour in all animals of that species."

    This comment is consistent with the "broad purpose" for paragraph (b) formulated by Slade LJ in Curtis v Betts but omitting the "one exception".

    126. In Gloster v Chief Constable of Greater Manchester Police, Pill LJ expressed agreement with Lloyd LJ in Breeden that section 2(2) was "not concerned with animals behaving in a perfectly normal way for animals of the species or sub-species" (p. 5 of the transcript). Gloster was a case in which a trained police dog, a German Shepherd, had bitten a well-meaning member of the public who was attempting to help the police, instead of the miscreant at whom he had been directed by his police-handler. The trial judge and the Court of Appeal declined to find that strict liability under section 2(2) was established. Pill LJ, adopting the single test construction of paragraph (b), held that the case did not fall within the paragraph because although a propensity to bite was not a characteristic normally found in German Shepherd dogs, it was a characteristic of the sub-species to respond to specific training and instruction. He said that the dog "acted as he was trained to act and in a way characteristic of the sub-species" (p. 6 of the transcript). In my respectful opinion, this reasoning cannot be right. A trained police dog is not a sub-species or variety of dog within the meaning of the section 11 definition. If biting at its handler's command is a characteristic of German Shepherds only after they have been trained to do so, it is not a normal characteristic of the sub-species.

    127. Hale LJ, the other member of the two person court, expressed doubts about Pill LJ's approach to paragraph (b). But she held that paragraph (a) was not satisfied, so concurred in the result.

    128. In my opinion, Gloster was a case in which, on the single test approach, paragraph (b) should have been held to be satisfied. The damage complained of was the bite. The likelihood of being chased and bitten was due to a characteristic of the police dog not normally found in German Shepherd dogs. There was, in my opinion, no more to be said about paragraph (b) than that.

    129. I have already referred to some of the results of applying a dual-test construction of paragraph (b) that imposes strict liability for damage caused by normal behaviour of animals that do not belong to a dangerous species. The alternative single test construction may also be regarded as producing in certain circumstances odd results. One of these oddities results from the definition of dangerous species. An animal cannot belong to a "dangerous species" if it belongs to a species commonly domesticated in the United Kingdom. But some such animals are, relatively speaking, animals that might reasonably be regarded as often dangerous. Bulls are one example. Stallions are another. I recall reading a comment by a horse expert that an angry stallion can be one of the most dangerous animals to have to face. If the single test construction of section 2(2)(b), suggested in paragraph 19, is right, there can never be strict liability for normal behaviour by domesticated animals of this character. If damage is caused by a charging bull, charging in circumstances in which it would be normal behaviour for a bull to charge, there would be no strict liability for the damage caused by the bull. The same would be true of damage caused by a stallion attacking in circumstances in which it would be normal for a stallion to attack. The same would be true of the bitch who bites a stranger attempting to handle her pups. Can this have been Parliament's intention?

    130. A clear answer to the question as to the proper construction of paragraph (b) cannot, in my opinion, be obtained from the actual language of the provision, nor from a perusal of Hansard, nor from examining the contents of the Law Commission Report of 1967 on which the 1971 Act was in part based. The answer depends upon identifying what Parliament appears to have been trying to achieve. It seems to me that Parliament was trying to draw a distinction between animals that in normal circumstances behaving normally are dangerous and those that in normal circumstances behaving normally are not. As to the former, they belong to a dangerous species and there was to be strict liability for damage; as to the latter they do not belong to a dangerous species and strict liability was to be limited to damage caused by the animal displaying abnormal characteristics that it was known by its keeper to possess. This seems to me to be a coherent policy. In respect of damage for which no strict liability was imposed, a remedy in negligence would always be available if the keeper of the animal had failed to exercise reasonable care to see that the animal did not cause damage. The keeper's knowledge of the circumstances in which and times at which the animal might be likely to become dangerous and cause damage would, of course, be highly relevant in determining the standard of care required to be observed by the keeper. A standard of care can, in appropriate circumstances, be placed so high as to require the person subject to it to become virtually an insurer against damage.

    131. Essentially, I am in respectful agreement with the approach of Lloyd LJ in Breeden v Lampard. I agree with his explanation of the function to be served by the concluding words of paragraph (b). I share the disbelief expressed by Oliver LJ at p. 12 of the transcript. A construction of paragraph (b) under which abnormal behaviour is a requisite for strict liability seems to me consistent with the statutory language, to promote the apparent scheme of the Act by confining strict liability under section 2(2) to cases where damage has been caused by animals displaying characteristics which are not normal for their species and to avoid the anomalies produced by the dual test construction. The construction, as Lloyd LJ noted, attributes the inclusion in paragraph (b) of the concluding words "or are not normally so found except at particular times or in particular circumstances" to the need to forestall attempts by defendants to escape the opening words, "the likelihood of the damage or of its being severe was due to characteristics of the animal which are not normally found in animals of the same species", by contending that all dogs will sometimes bite, all horses will sometimes kick, all bulls will sometimes charge etc. If, as I think it was, the intention of paragraph (b) was to require the likelihood of the damage, or its severity, to be attributable to particular, individual characteristics of the delinquent animal not shared by others of its species, some additional words were necessary to prevent escape attempts of the sort described. The language might have been better chosen and have avoided the ambiguity that has caused the problem. But the suggested construction is, in my opinion, consistent with the statutory language actually used by Parliament and is one that I think your Lordships can and should adopt.

    132. In the instant case, the three horses bolted and burst through their fence because something or other had terrified them. Bolting when terrified is a characteristic of all horses. The findings of fact at trial were that the behaviour of these horses was not due to characteristics not normally found in horses. The concluding words of paragraph (b) do not impose a second alternative test but merely amplify abnormality if abnormality has been found. I would, accordingly, allow the appeal.

LORD WALKER OF GESTINGTHORPE

My Lords,

    133. Many difficulties have arisen in interpreting and applying section 2(2) of the Animals Act 1971 ("the Act"). That subsection deals with the circumstances in which strict liability is imposed (subject to defences under section 5) on the keeper of an animal which does not belong to a dangerous species (as defined in sections 6(2) and 11) for damage caused by the animal. Hale LJ has (in paragraph 18 of her judgment) collected an anthology of critical observations on section 2(2) made by some distinguished judges, and it is not necessary to repeat them. But it may be worth reflecting on why section 2(2) has given rise to so many difficulties.

    134. It is not necessary to go far into the old common law rules which imposed strict liability for wild animals (animals ferae naturae) or for tame or domesticated animals with a known vicious propensity (the scienter basis of liability). The old rules were both questionable in their foundations and uncertain in their limits. That appears from two cases decided not very long before the Act, Behrens v Bertram Mills Circus Limited [1957] 2 QB 1 (the case of the trained Burmese elephant which was more docile than many horses until harassed by a small dog) and Fitzgerald v Cooke Bourne (Farms) Ltd [1964] 1 QB 249 (the case of an unbroken filly in a field crossed by a public footpath). The Goddard Committee (which reported in 1953) proposed to abolish strict liability for damage caused by animals, but the Law Commission in its Report on Liability for Animals (published in 1967 as one of the Law Commission's earliest reports) took a different view. The Law Commission recommended that the principle of strict liability should not be abolished, but should be modified and simplified. It is clear that in enacting the Act, Parliament was (in the most general terms) following the Law Commission's recommendations to retain the principle in a modified form. It is unfortunately far from clear that Parliament achieved the objective of simplification.

    135. Part of the problem is that section 2 of the Act is expressed in very general terms. It is notable that the Law Commission inquired into the prevalence of particular types of damage caused by animals. Its report contains some detailed statistics about road accidents in which animals were involved. But in section 2 Parliament has not chosen to identify or make specific provision for the varying circumstances in which animals do most commonly cause damage. In practice section 2 (1) has a very narrow scope, being almost entirely limited to incidents in (or following escapes from) zoos or circuses. Section 2(2) has to cover the whole range of incidents involving animals of species classified as non-dangerous (which I will call domesticated animals, although that is not an entirely accurate term). That range includes (i) physical injury to humans by biting (especially by dogs) or kicking or knocking down (especially by horses); (ii) injuries caused to livestock (such as a dog worrying a neighbour's sheep, or a cat killing a neighbour's chickens); (iii) road traffic accidents, especially those caused by animals straying on the highway; (iv) damage caused by livestock getting out onto neighbouring land and destroying crops or gardens; and (v) injury or damage caused by the spread of animal infection or by the smell or noise of animals (a class which shades off into cases normally classified as nuisance). So section 2(2) has a lot of work to do. It is expressed in general, abstract terms and it has to be applied to a wide range of disparate incidents.

    136. Other sections of the Act do contain more specific provisions. The case of livestock trespassing on private land is covered by section 4, and there is a special provision as to guard dogs injuring trespassers (section 5(3)). But the only special provision made for animals straying on the highway is the abolition by section 8 (subject to qualifications in section 8(2)) of the old common law rule which gave immunity (see Searle v Wallbank [1947] AC 341). It has not been contended in your Lordships' House (although it was contended at first instance) that section 8 has the effect of excluding possible liability under section 2(2).

    137. Section 2(2) must therefore be treated as capable of applying (one way or another) to cases of horses straying onto a highway and causing an accident, as well as to cases (such as Cummings v Granger [1977] QB 397 and Curtis v Betts [1990] 1 WLR 459) where humans have been injured by being bitten by large dogs. Many of the epithets used in the cases relating to dog bites (such as fierce, ferocious and vicious) are not apposite to describe a horse sent into a state of panic by some unknown cause. But it cannot be doubted that for a riderless horse to be on the highway in such a state is a danger to other road users, even though it is (in its state of panic) acting in an entirely natural way. If the Court of Appeal was right the Act has in this respect extended the possible scope of strict liability for domesticated animals (while narrowing the class of dangerous species by the definition in section 6(2)).

    138. After these general comments I come to the particular linguistic difficulties presented by section 2(2). One is the meaning of the important term "characteristics" used in paragraphs (b) and (c) of section 2(2), but not defined in the Act. The context makes clear that the expression cannot mean something buried in an animal's psyche (as Devlin J said in Behrens at page 18, it is not practical to introduce conceptions of mens rea and malevolence in relation to animals). It must refer to character or disposition as evinced by overt behaviour—for instance, a dog which had the habit of attacking people who were carrying bags (Kite v Napp, Times Newspaper 1 June 1982). The distinction between "permanent" and "temporary" characteristics drawn by Stuart-Smith LJ in Curtis v Betts [1990] 1 WLR 459, 469, is useful but must be treated with some caution: all dangerous characteristics are likely to be more or less permanent but they may show themselves either frequently and randomly (as with the unreliable horse in Wallace v Newton [1982] 1 WLR 375), or under a stimulus peculiar to the particular animal (such as bag-carrying in Kite v Napp), or under some internal or external stimulus (such as the animal's hormones or a perceived challenge to its territory) which can be expected to produce similar behaviour in most animals of its species.

    139. That is the point to which the words "at particular times or in particular circumstances" are directed, but there is force in the observation made by the trial judge, in his careful judgment, that one can always find particularity attaching to any time or to any circumstance. I consider that Mr Sharp QC (for the respondent) must be right in suggesting that predictability (of how animals of the same species react to a particular stimulus or situation) is one of the indicia of characteristic behaviour which falls within the second limb of section 2 (2)(b).

    140. That leads to the central problem on this appeal. It is agreed that section 2(2)(b) contains two limbs, linked by the word "or". The second limb contains what is akin to a double negative ("not …except …") and this (coupled with the cumbersome words at the beginning of paragraph (b), the feature which has so far attracted most of the adverse judicial comment) makes it difficult to see what paragraph (b) as a whole is getting at. The cumbersome words at the beginning appear to me to reflect the simple proposition (familiar from the law of negligence: see for instance Paris v Stepney Borough Council [1951] AC 367) that risk is a product of two factors, the likelihood of injury and the severity of the possible injury. So the sub-section could be set out in a simplified form (using the abbreviation "risk" and some other simplifications) as follows:

"The [risk] was due to characteristics of the [horse] which are not normally found in [horses] or are not normally … found [in horses] except [on particular occasions]".

    141. If paragraph (b) is simplified in this way, it is easier to see that there are two possible interpretations of the second limb. Each is permissible (although not necessarily equally acceptable) as a matter of language. Which is to be preferred depends on the legislative context and purpose, and in particular, on what appears to be the essential purpose of the second limb as a whole. This can be illustrated by the example (based on Barnes v Lucille Limited (1907) 96 LT 680 and discussed both by the Law Commission and in later authorities) of the bitch which acts fiercely and bites in defence of her pups. Suppose that a labrador bitch (which is not nursing pups and is not subjected to any other provocation) bites a pedestrian in the park. That would on the face of things be abnormal behaviour for a labrador, and the first limb of paragraph (b) would apply. The only function of the second limb (one argument goes) is to forestall the owner's excuse, "but all labrador bitches have a propensity to bite sometimes" in a case where that excuse cannot, on the facts, make any difference.

    142. The competing explanation of the second limb is that it adds a further possible head of liability where the particular circumstances are actually present (in the example, where the bitch is nursing pups). In such a case the animal's normal behaviour in abnormal circumstances is equated with a more vicious dog's abnormal behaviour in normal circumstances. Either is to be treated as introducing the element of abnormal, dangerous behaviour which goes towards the establishment of strict liability, if the other elements (in paragraphs (a) and (c) of section 2(2)) are also present.

    143. That is the explanation which was preferred by the Court of Appeal in Cummings v Granger and Curtis v Betts. In the latter case, Slade LJ said at page 464,

"The broad purpose of requirement (b), as I read it, is to ensure that, even in a case falling within requirement (a), the defendant, subject to one exception, will still escape liability if, on the particular facts, the likelihood of damage was attributable to potentially dangerous characteristics of the animal which are normally found in animals of the same species. The one exception is this. The mere fact that a particular animal shared its potentially dangerous characteristics with other animals of the same species will not preclude the satisfaction of requirement (b) if on the particular facts the likelihood of damage was attributable to characteristics normally found in animals of the same species at times or in circumstances corresponding with those in which the damage actually occurred".

    144. Similarly Stuart-Smith LJ said of the two limbs of section 2(2)(b), after referring to permanent and temporary characteristics, at page 469,

"Dogs are not normally fierce or prone to attack humans; a dog which has a propensity to do this at all times and in all places and without discrimination as to persons would clearly fall within the first limb. One that is only aggressive in particular circumstances, for example, when guarding its territory or, if a bitch, when it has a litter of pups, will come within the second limb".

    145. The weight of authority favours the view taken by the Court of Appeal in Curtis v Betts (with dicta of two members of the Court of Appeal in Breeden v Lampard going the other way). But Mr Lissack QC (for the appellants) has strenuously argued that the current of authority is wrong, because (contrary to Parliament's general purpose) it treats normal animal behaviour as if it were abnormal. Echoing the Law Commission report (paragraph 15 (ii)) he submitted that your Lordships should redirect the law so that it can reflect the common experience of everyday life.

    Breeden v Lampard was the case in which one mounted follower of the hunt had her leg broken when she was kicked by the horse of another member of the hunt. The field had just changed direction and several horses were in close proximity to one another. The defendant does not seem to have run the defence of voluntary acceptance of risk provided by section 5(2) of the Act. The offending horse was wearing a red patch on its tail but the trial judge found that it was not a kicker and that its rider had not been negligent. He dismissed the claim.

    147. In the Court of Appeal Sir George Waller (who gave the first judgment) regarded the case as not within either the first limb or the second limb of section 2(2)(b). In any event the defendant was not aware of any exceptional characteristics in her horse. Lloyd LJ went into the matter rather more fully, summarising the appellant's argument on the second limb as follows:-

"the argument went like this: Any horse is liable to kick out when approached too closely, or too suddenly, from behind. That is a characteristic of all horses, or at any rate of all young horses, in those circumstances. Those circumstances are particular circumstances within the meaning of section 2(2)(b) of the Act, and therefore the defendant is under strict liability by virtue of that subsection. The short answer to that argument is that which has been given by my Lord; this defendant did not know of that particular characteristic in relation to her horse. The judge accepted her evidence without reservation; it follows that even if the provisions of section 2(2)(b) are satisfied here, the provisions of section 2(2)(c) are not".

    148. Lloyd LJ then expressed doubt as to whether the second limb could have applied anyway. After some remarks about the old law and the Law Commission's report he referred to cases (mentioned in North, the Modern Law of Animals (1972) page 50) of strict liability for injury caused "by a vice natural to the species" of a domesticated animal (see for instance Buckle v Holmes [1926] 2 KB 125, the case of a cat which killed a neighbour's pigeons and bantams). Lloyd LJ then said:

"But all that has now been swept away by section 2(2)(b) of the new Act. The essential condition for liability now is that the characteristic which is known to the owner must be a characteristic which is abnormal for the species. I cannot myself see why, if the old law has been swept away, Parliament should have retained by way of exception the effect of the decision in Barnes v Lucille Limited if indeed the effect is as Mr Nicholl contends. If liability is based on the possession of some abnormal characteristic known to the owner, then I cannot see any sense in imposing liability when the animal is behaving in a perfectly normal way for all animals of that species in those circumstances, even though it would not be normal for those animals to behave in that way in other circumstances, for example, a bitch with pups or a horse kicking out when approached too suddenly, or too closely, from behind".

    He then suggested an alternative explanation for the second limb (that favoured by the appellants in your Lordships' House). Oliver LJ gave a short judgment stating that he shared Lloyd LJ's puzzlement and observing, "I cannot believe that Parliament intended to impose liability for what is essentially normal behaviour in all animals of that species".

    149. The Law Commission might have been surprised at the suggestion that the old law was being swept away (rather than retained with modifications). But before going further into the Law Commission report or the parliamentary material which your Lordships were asked to consider, I prefer to return to the text of the Act. The skilled eye of the parliamentary draftsman can hardly have failed to spot the difficulty, and yet the language of section 2 (2) does not avoid ambiguity. Section 2(2) as a whole packs several complex ideas into a single sentence, and the draftsman may have felt that there was no room for any more subordinate clauses to be included. Had he not been constrained in that way he might have expressed the first alternative meaning (favoured by the appellants) on the following lines:

"the risk is due to characteristics of the animal which (i) are abnormal in its species or (ii) are normal in the species but only at particular times or in particular circumstances (and the danger is not caused at such a time or in such circumstances)".

    Conversely he might have expressed the second alternative (favoured by the respondent) as follows:

"The risk is due to characteristics of the animal which (i) are abnormal in its species or (ii) are normal in the species but only at particular times or in particular circumstances (and the damage is caused at such a time or in such circumstances).

    150. On either view the first limb covers wholly abnormal behaviour. The respondent's interpretation of the second limb expressly extends the scope of possible liability to behaviour which, although generally abnormal, is normal for the species in particular circumstances which were those of the incident. The appellants' interpretation expressly excludes such semi-normal, semi-abnormal behaviour simply in order to give a fuller explanation of what the first limb means in a case where the incident occurred in circumstances which were not the sort of "particular circumstances" envisaged.

    151. The appellants' interpretation appears to me to be less likely as a matter of language. For one thing, if that were the intended meaning, it could have been more simply expressed by a parenthesis—"with the possible exception of [or apart from] abnormal behaviour in particular circumstances"—rather than by an apparently free-standing alternative. As Lord Diplock said (in his dissenting speech in Carver v Duncan [1985] AC 1082, 1117-8) Parliament does not normally use the word "or" to mean "that is to say". For another thing, the appellants' interpretation, by expressly excluding semi-normal, semi-abnormal behaviour in a case where that behaviour would not be expected, raises but does not answer the highly pertinent question of what is to be the position if the circumstances are such that the behaviour is to be expected. That would be a surprising way of framing legislation which is meant to simplify and clarify the law. These considerations make the respondent's interpretation significantly easier as a matter of language, but not to my mind so much more probable as to be determinative of the issue.

    152. The course of argument before your Lordships identified two general arguments in favour of the appellants' proposed construction of the second limb of section 2(2)(b). Both are based on the anomalous (or even absurd) results said to follow from the alternative construction:

It would be anomalous or absurd if the keeper of an animal is not strictly liable for damage caused by a domesticated animal which (in common with others of its species) evinces dangerous behavioural characteristics all the time, but is strictly liable for damage caused by a domesticated animal which (in common with others of its species) behaves in a dangerous way only in particular circumstances.

It would be anomalous or absurd, if strict liability may be imposed for behaviour which is in some sense entirely normal for an animal of the species, to make the keeper's liability depend on his or her knowledge of something which is likely to be common knowledge among those who keep animals of the species.

    153. These points call for serious consideration. They both share a common foundation in scepticism (vividly expressed by Lloyd and Oliver LJJ in Breeden v Lampard) that Parliament cannot have intended to push out the boundaries of strict liability so as to extend to normal behaviour on the part of an animal of a domesticated species. However Parliament has (by its chosen definition of dangerous species) drawn the line so that large, strong animals such as bulls, cows and horses, and potentially savage animals such as mastiffs and rottweilers, are classified as not belonging to a dangerous species. Bulls may or may not be potentially dangerous all the time (this House in its legislative capacity expressed widely differing views on the subject in the debate in 1970 on the second reading of the Animals Bill) but they are certainly dangerous in particular circumstances. On a smaller scale the same is true of domestic cats, which are instinctive and ruthless killers of birds; it would have been little comfort to the owner of the dead pigeons and bantams in Buckle v Holmes to be told that the cat was behaving in an entirely natural way. As Devlin J said in Behrens v Bertram Mills Circus Limited, the court must avoid misplaced notions of an animal's guilty mind. It must also, I think, avoid any notion (reminiscent of Jean-Jacques Rousseau) that an animal's natural behaviour must be somehow innocent.

    154. I am not persuaded that the two suggested anomalies, either separately or together, are decisive of the issue of construction. For a dog to jump up and bite in defence of its territory, or for a horse to kick out if approached from behind, may be normal behaviour for the species, but it is abnormal behaviour, at a higher level of generality, for a species which is supposedly tame and domesticated. Moreover although these traits will usually be known to any knowledgeable dog-owner or horse-owner, that may not be so (as the trial judge's finding in Breeden v Lampard illustrates) and it certainly cannot be assumed that they would be known to every member of the general public.

    155. In my view the crux of the matter is this. Both sides agree that Parliament intended to impose strict liability only for animals which are (in some sense) dangerous. Subsections (1) and (2) of section 2 mark the first subdivision which Parliament has made in identifying one (very limited) class of dangerous animals. This rather crude subdivision has contributed to the difficulties which have arisen, since it implies (but does not clearly spell out) that entirely normal behaviour of an animal of a non-dangerous species can never give rise to strict liability (this is the basis of the first anomaly relied on by the appellants). Domesticated animals are to be the subject of strict liability only if their behavioural characteristics are (in some sense) abnormal (and so dangerous). Did Parliament contemplate that the generality of animals in a domesticated species might in some circumstances show dangerous behavioural characteristics so as to be liable to be treated, in those circumstances, as dangerous? Or is there a presumption underlying the Act (and providing guidance as to the correct construction of section 2) that an animal of a domesticated species behaving in a way that is (in particular circumstances) normal and natural for its species cannot be treated as dangerous?

    156. In my view the scheme and language of the Act do not yield any such underlying presumption. I consider that the respondent's proposed construction of the second limb of section 2(2)(b) is more natural as a matter of language, and that it is not inconsistent with Parliament's general intention to impose strict liability only for animals known to present special dangers. The suggested anomalies, although far from insignificant, could be matched by comparable anomalies arising from the alternative construction. Moreover the respondent's proposed construction is in my view closer to what Mr Lissack QC (echoing the Law Commission) referred to as the common experience of everyday life.

    157. It is common knowledge (and was known to the appellants in this case) that horses, if exposed to a very frightening stimulus, will panic and stampede, knocking down obstacles in their path (in this case an electric fence, a post and barbed wire fence behind that, and then high undergrowth) and may continue their flight for a considerable distance. Horses loose in that state, either by day or by night, are an obvious danger on a road carrying fast-moving traffic. The appellants knew these facts; they could decide whether to run the unavoidable risks involved in keeping horses; they could decide whether or not to insure against those risks. Although I feel sympathy for the appellants, who were held not to have been negligent in the fencing of the field, I see nothing unjust or unreasonable in the appellants having to bear the loss resulting from their horses' escape rather than the respondent (who suffered very serious and painful injuries in the accident, although he was wearing a seatbelt and slowed down as soon as he saw the first horse in his headlights).

    158. The Law Commission report provides useful background material to the Act but is to my mind of little or no assistance on the crucial issue of construction. Two paragraphs of the report (paragraph 18(i) and the summary at paragraph 91(iv)) appear to favour the respondent but the draft Bill prepared by the Law Commission did not cover the point one way or the other (it referred to "characteristics" without indicating whether they had to be normal or abnormal).

    159. Your Lordships have been asked to consider (and have considered on a provisional basis) a quantity of parliamentary material. I share what I understand to be your Lordships' reluctance to extend the clear guidelines set out in the speech of Lord Browne-Wilkinson in Pepper v Hart [1993] AC 593, 640. The only ministerial statement which seems to me to come close to those guidelines is the statement made by Lord Hailsham LC in the House of Lords on 12 November 1970, on the question whether clause 2 should stand as part of the Bill. At the end of a discussion initiated by Lord Kilbracken the Lord Chancellor said

"I have tried to give what guidance I can to the Committee on the subject of bulls. They are, I would apprehend, normally domesticated in the British Isles. That is the first proposition. Any damage which the ordinary bull is liable for is damage which would be of the kind normally to be expected of the bovine species. If, however, a bull is known to have vicious characteristics, or if it be that there are particular times and conditions in which abnormally vicious characteristics appear in all bulls, the keeper of the bull would be liable".

    160. The first and second conditions stated by Lord Browne-Wilkinson (the obscurity of the legislation, and the statement being that of a minister promoting the Bill) are satisfied. I feel some doubt as to the third (that the statement relied on is clear) since in the passage quoted the Lord Chancellor's penultimate sentence (if directed to strict liability rather than negligence) seems to me, with great respect, to be mistaken. Nevertheless the last sentence clearly favours the respondent's position. The passage provides some support for my conclusion, but I do not place much weight on it.

    161. On the other principal issue in the appeal, the issue of causation, I see some force in the submission of Mr Lissack QC that it was illogical for Hale LJ to deal with this issue first, before she had dealt with the main issue of construction. However the essential point is that in order to recover the claimant had to show that the damage which he had suffered was caused, not merely by the horses escaping and being on the main road, but by the characteristics which are capable of founding strict liability under section 2(2)—in short, a frightened horse's propensity to bolt, to continue to flee, and to ignore obstacles in its path.

    162. The trial judge (following the Court of Appeal in Jaundrill v Gillett [16 January 1996] ) thought that the damage was caused by the presence of the horses on the highway, rather than by any relevant characteristic. Hale LJ and the other members of the Court of Appeal took a different view. Hale LJ said at para. 16,

"In this case, however, it is indeed difficult to conclude that it was anything other than the particular characteristics of these horses once they had been terrified which led to their escape and to this accident taking place. They were still not behaving in the ordinary way in which they would behave when taken on the road. One witness referred to them bolting; another to them trotting across the road in front of the vehicles; they crashed into the vehicles rather than the other way about. It is precisely because they were behaving in the unusual way caused by their panic that the accident took place".