CE Heath Underwriting & Insurance (Aust) Pty Ltd v Edwards Dunlop & Co Ltd

The critical question on the appeal is whether, in circumstances where the policy was for an initial period of twelve months and was subsequently renewed for successive annual periods, the phrase "the Policy Period" in Section 1 should be construed as referring to the particular annual period of insurance in which a loss occurred or to the overall period in which "the Policy" remained in operation, that is to say, the total of the original annual term and of the subsequent terms for which "the Policy" was renewed. The learned trial judge (Brownie J.) and Meagher J.A., who dissented in the New South Wales Court of Appeal, preferred the construction for which the Insurer contends, namely, that the phrase "the Policy Period" refers only to the particular annual term of insurance in which a loss occurs. The majority of the Court of Appeal (Clarke J.A. and Hope A.J.A.) preferred the construction for which the Insured contends, namely, that the phrase refers to the overall period in which "the Policy" remained in operation. It can be said at once that I agree with the majority of the Court of Appeal. Accordingly, I would dismiss the appeal.

The detailed facts are set out in the judgments in the courts below which have been reported [1] and in the judgment of Dawson, Toohey and McHugh JJ. in this Court. It is unnecessary that I repeat them. Nor, in a context where the outcome of the appeal turns upon the construction of a particular phrase in a particular policy of insurance and where the conclusion which I have reached is a dissenting one, is it necessary that I do more than indicate that, subject to the comments which follow, I am in general agreement with the judgment of Clarke J.A. (in which Hope A.J.A. concurred) in the Court of Appeal.

1. See Edwards Dunlop & Co. Ltd. v C.E. Heath Underwriting & Insurance (Australia) Pty Ltd (1989), 6 A.N.Z. Insurance Cases 76,238 (Brownie J.); (1991), 6 A.N.Z. Insurance Cases 77,066 (Court of Appeal).

As has been seen, the function of the Policy was to cover the Insured against any loss of money or other property sustained through "fraudulent or dishonest act or acts" committed by its employees. It is commonplace for losses sustained by the fraudulent or dishonest conduct of an employee to remain successfully concealed by the employee concerned for a considerable period of time. Indeed, in a case of a continuing course of fraud or dishonesty, it may be very difficult to ascertain precisely when the relevant "loss of money or property" was sustained. In these circumstances, the construction for which the Insurer contends would result in a significant gap in the cover against relevant losses notwithstanding that they were sustained and discovered within the overall period of insurance with the Insurer. A loss would be excluded from the Insurer's continuing cover if it remained concealed for between one and two years, the precise period of concealment necessary to give rise to exclusion depending upon whether the loss occurred at the end or beginning of the then current year of the policy. Such a gap in cover is something which a prudent insured could be expected to anticipate and guard against when changing from one insurer to another. In my view, however, it is not something which would ordinarily be expected in the cover under a policy which is renewed from year to year with the same insurer. In that regard, it is relevant to note that no such gap existed in the earlier "Industrial Special Risks" policy between the Insurer and the Insured which the Blanket Fidelity Policy partially replaced. Under that earlier policy, the period of insurance was defined to include any "renewal of [the] policy" and the relevant restriction applied only to exclude a loss resulting from employee fraud or dishonesty if the loss was not discovered "within twelve months of the termination of [the] policy" (emphasis added).

As the difference of opinion between the judges in the courts below [2] indicates, there is real ambiguity about the meaning of the phrase "the Policy Period" in Section 1 when that phrase is construed in the context of the Policy as a whole. Indeed, Meagher J.A., who favoured the Insurer's construction, robustly commented that the difficulty of the question of construction "chiefly arises because the policies, and other documents, emanating from the insurer could not be more perplexing if they had been specifically drafted in order to generate ambiguity" [3] . I would not, for my part, put the matter quite as strongly as that. It does, however, seem to me to be plain that an ordinary commercial person would construe the phrase "the Policy Period" in a policy of insurance which was renewed from year to year as prima facie referring to the overall period during which the cover provided by the policy continued. That is certainly the prima facie meaning which a lawyer would be expected to give to the phrase since "policy of insurance" is ordinarily understood by lawyers as referring not to a particular annual contract but to the written terms and conditions of the cover [4] . Indeed, the word "Policy" is clearly used in that sense by the Insurer in some other parts of the Policy [5] . It is true that the contrary construction for which the Insurer contends obtains some support from a detailed analysis of the relationship between different provisions of the Policy. It is scarcely to be expected, however, that it would occur to an insured person that he or she should pursue that kind of detailed analysis for the purpose of ascertaining whether the prima facie meaning of a phrase such as "the Policy Period" has been displaced by obscure contextual considerations.

1. See, (1989) 6 A.N.Z. Insurance Cases, pp. 76,239-76,240, per Brownie J.; (1991) 6 A.N.Z. Insurance Cases, p. 77,071, per Clarke J.A.; p. 77,073, per Meagher J.A. 2. (1991) 6 A.N.Z. Insurance Cases, at p. 77,073. 3. See, e.g., Halsbury's Laws of England, 4th ed., vol. 25, pars 405-406; Davies v Sprod , [1943] S.A.S.R. 53, at pp. 57-58. 4. See, e.g., Section 9 which speaks "of the number of years this Policy shall continue in force."

In these circumstances, the construction for which the Insured contends and which I consider to be the preferable one derives strong support from the application of the settled rule of construction that, if there be difficulty or obscurity of meaning, the words of an instrument are to be interpreted more strongly [6] against him or her whose instrument it is. That rule of construction has been subjected to some distinguished criticism [7] . It is, however, deeply rooted in the common law [8] and fully justified by considerations of justice and common sense [9] . It is, "particularly applicable to contracts of insurance" [10] . As Willes J. said more than one hundred and twenty-five years ago in Fitton v The Accidental Death Insurance Co. [11] : "[I]t is extremely important with reference to insurance, that there should be a tendency rather to hold for the assured than for the company, where any ambiguity arises upon the face of the policy." Put differently, the courts should avoid a construction of a policy of insurance which has the effect that the insured, having paid a premium to be indemnified against loss caused by a particular kind of occurrence, is subjected to a new risk of not being insured against a loss of that kind by reason of the ambiguity or obscurity of the language which the insurer has seen fit to use in its policy.

1. More often "strongest" or, in some old texts, "strongliest". 2. See, in particular, Taylor v Corporation of St. Helens (1877), 6 Ch. D. 264, at pp. 270-271, per Jessel M.R. 3. See, e.g., Co. Litt. 36(a); Anderson v Fitzgerald (1853), 4 H.L.C. 484, at pp. 507, 510 [10 E.R. 551, at pp. 560, 561]; Fowkes v Manchester and London Life Assurance and Loan Association (1863), 3 B. & S. 917, at pp. 929-930 [122 E.R. 343, at pp. 347-348]; Western Australian Bank v Royal Insurance Co. (1908), 5 C.L.R. 533, at pp. 559, 567-568; Wilshire v Guardian Assurance Co. Ltd. (1912), 15 C.L.R. 516, at p. 528; Guardian Assurance Co. v Condogianis (1919), 26 C.L.R. 231, at pp. 235, 239, 242-244; Condogianis v Guardian Assurance Co. Ltd. , [1921] 2 A.C. 125, at p. 130; Maye v Colonial Mutual Life Assurance Society Ltd. (1924), 35 C.L.R. 14, at pp. 22, 27; Minucoe v London & Liverpool & Globe Insurance Co. Ltd. (1925), 36 C.L.R. 513, at p. 523; Halford v Price (1960), 105 C.L.R. 23, at p. 30; Steadfast Insurance Co. Ltd. v F. & B. Trading Co. Pty Ltd (1971), 125 C.L.R. 578, at p. 581. 4. See Bacon, The Elements of the Common Lawes of England, in Classics of English Legal Theory in the Modern Era (1978), Maxim 3 at p. 11: "a rule drawn out of the depth of reason." 5. Australian Widows' Fund Life Assurance Society Ltd. v National Mutual Life Association of Australasia Ltd. (1912), 14 C.L.R. 141, at p. 155, per Griffith C.J. See also Halford v Price (1960), 105 C.L.R., at p. 34; Life Association of Scotland v Forster (1873), 11 M. 351, at p. 369 [11 Ct. of Sess. 371, at p. 387]; American Surety Co. v Pauly [No. 1] (1897), 170 U.S. 133, at p. 144; Mutual Insurance Co. v Hurni Co. (1923), 263 U.S. 167, at p. 174. 6. (1864) 17 C.B. (N.S.) 122, at pp. 134-135 [144 E.R. 50, at p. 55]. Dawson, Toohey and McHugh JJ.

The respondent took out an insurance policy with the appellant underwriters which was described as an "Industrial Special Risks Policy". Amongst other things it insured the respondent against "loss of the property insured resulting from any fraudulent or dishonest act(s) committed by any employee of the Insured" occurring during the period specified in the schedule (the "period of insurance"). There were two material limitations in respect of this type of loss. First, the amount of the appellant's liability was limited to $500,000 "in respect of any one person acting alone or in collusion with others, and in the aggregate during the period of insurance". Secondly, the loss covered did not include "any loss unless discovered within twelve months of the termination of [the] policy".

Under the heading "period of insurance" in the schedule were the following clauses:

(i) This policy attaches and is effective from 4.00 p.m. local time, 31st May, 1983 and terminates at 4.00 p.m. local time, 30th April 1984

(ii) Any subsequent period for which the Insured shall pay or agree to pay and [the appellant] agree to accept a premium for the renewal of this policy.

The respondent and the appellant renewed the Industrial Special Risks Policy for the period from 30 April 1984 to 30 April 1985.

The respondent subsequently took out another policy, a "Blanket Fidelity Policy", with the appellant. Under that policy the appellant undertook to indemnify the respondent "against any loss of money or other property which the Insured shall sustain through any fraudulent or dishonest act or acts committed by any of the [Insured's employees], acting alone or in collusion with others". The policy comprised sections headed respectively "Insuring Agreement", "General Agreements", "Conditions and Limitations" and "Declarations". General Agreement C provided:

If the coverage of this Policy is substituted for any prior policy of insurance carried by the Insured , which prior policy is terminated, cancelled or allowed to expire as of the time of such substitution, the [appellant] agrees that this Policy applies to loss which is discovered as provided in Section 1 of the Conditions and Limitations and which would have been recoverable by the Insured under such prior policy except for the fact that the time within which to discover loss thereunder had expired; provided:

1. the indemnity afforded by this General Agreement C shall be a part of and not in addition to the amount of insurance afforded by this Policy;

2. such loss would have been covered under this Policy had this Policy with its agreements, limitations and conditions as of the time of such substitution been in force when the acts or defaults causing such loss were committed; and

3. recovery under this Policy on account of such loss shall in no event exceed the amount which would have been recoverable under this Policy in the amount for which it is written as of the time of such substitution, had this Policy been in force when such acts or defaults were committed, or the amount which would have been recoverable under such prior policy had such prior policy continued in force until the discovery of such loss, if the latter amount be smaller.

Section 1 of the Conditions and Limitations provided:

Loss is covered under this Policy only if discovered not later than one year from the end of the Policy Period. Subject to General Agreement C, this Policy applies only to loss sustained by the Insured through fraudulent or dishonest acts committed during the Policy Period by any of the [Insured's employees] engaged in the regular service of the Insured within the Commonwealth of Australia.

The "Policy Period" was defined in Item 2 of the Declarations as "12 months from noon on 30th April, 1985". Item 3 of the Declarations limited the appellant's liability under the policy to $500,000 "any one employee and in all". Item 5 of the Declarations provided that acceptance of the Blanket Fidelity Policy by the respondent constituted notice to the appellant terminating or cancelling the Industrial Special Risks Policy, such termination or cancellation being effective as of the time the Blanket Fidelity Policy became effective. It is common ground that the Blanket Fidelity Policy was effective from 30 April 1985.

The Blanket Fidelity Policy was renewed for the years ending 30 April 1987, 30 April 1988 and 30 April 1989 subject to the terms and conditions of the policy. In the case of the first renewal the sum insured was $500,000 and in the case of the later renewals it was $1,500,000. Each renewal was expressed to be "any one employee/all employees any one period of insurance" (or words to similar effect).

Between 20 January 1982 and 29 June 1988 one of the respondent's employees misappropriated $695,450.62. The respondent discovered the misappropriations in July or August 1988. There is no dispute that the appellant was not liable to indemnify the respondent against the losses which occurred before 31 May 1983. This is because there was no relevant policy of insurance in existence before that time.

On the other hand, there is no dispute that the appellant was liable to indemnify the respondent in respect of the losses incurred after 30 April 1987 because those losses were clearly covered by the Blanket Fidelity Policy.

What is in dispute is the liability of the appellant under the Blanket Fidelity Policy to indemnify the respondent against the losses incurred between 31 May 1983 and 30 April 1987. The appellant's liability for losses incurred during that period depends upon the proper meaning to be attributed to the expression "Policy Period" in the Blanket Fidelity Policy. The respondent contends that the Policy Period was extended from time to time by the renewals which took place, so that there was one Policy Period which extended from 30 April 1985 to 30 April 1989. Upon this construction of the policy, the appellant would be liable for losses occurring during that period provided that the losses were discovered not later than 30 April 1990. Not only this, but on the substitution of the Blanket Fidelity Policy for the Industrial Special Risks Policy, the losses which would have been recoverable under the Industrial Special Risks Policy but for the failure to discover them within the time allowed by that policy would be recoverable under the Blanket Fidelity Policy provided they were discovered not later than 30 April 1990.

The appellant, on the other hand, contends that there were successive Policy Periods of twelve months each under the Blanket Fidelity Policy. Thus, in relation to the first year during which that policy was in force, the Policy Period was that year and, in relation to each of the succeeding years, the yearly period covered by each renewal. Accordingly the appellant argues that, so far as the losses incurred during the currency of the Industrial Special Risks Policy are concerned, they were not discovered within one year of the end of the first Policy Period of the Blanket Fidelity Policy, that is to say, before 30 April 1987, and so were not recoverable under that policy. Nor, upon this argument, were the losses incurred between 30 April 1985 and 30 April 1987 covered by the Blanket Fidelity Policy because they were not discovered within one year of the relevant Policy Period during which they were incurred, discovery of the losses not having taken place until July or August 1988.

In the course of argument considerable attention was devoted to whether the successive renewals of the Blanket Fidelity Policy resulted in a series of new contracts being entered into by the respondent and the appellant. The distinction between the renewal of a policy and the extension of a policy was expressed in the following terms by Mayo J. in Re Kerr [12] :

Strictly, a "renewal" is descriptive of a repetition of the whole arrangement by substituting the like agreement in place of that previously subsisting, to be operative over a new period, whereas an "extension" betokens a prolongation of the subsisting contract by the exercise of a power reserved thereby to vary one of its provisions, that is, by enlarging the period. Upon a renewal similar rights revest A contract reserving continuous rights of renewal will, if these be exercised, lead to succeeding contracts in a series, the identity of each contract [being] separate and distinct. On the other hand, the exercise of the right of extension augments the length of time over which the contract operates, without changing its identity. 1. [1943] S.A.S.R. 8, at p. 16.

Whether there is a renewal or an extension of an insurance policy is a question of construction [13] , the term "renewal" often being used to refer to both "renewal" and "extension" in the sense that those words are used above. It is, however, well established that, where a policy is renewable only by mutual consent (i.e. not as of right), the renewal results in a fresh contract rather than the extension of an existing contract [14] . Of course, a policy may expressly stipulate that it is not to continue in force beyond the period of insurance, unless renewed by mutual consent [15] . And where a policy, such as the ordinary form of life policy, expressly provides for continuation beyond the specified period of insurance unless a particular event, such as the non-payment of the premium, takes place, the renewal is an extension of the original contract [16] . But where a policy is silent on the question of renewal, renewal of it will generally constitute a new contract [17] .

1. See, e.g., Government Insurance Office (N.S.W.) v Kimmedy (1988), 5 A.N.Z. Insurance Cases 75,539, at p. 75,541. 2. See Re Kerr, [1943] S.A.S.R., at p. 15; Halsbury's Laws of England, 4th ed., vol. 25, par. 494. 3. See, e.g., Stokell v Heywood , [1897] 1 Ch. 459. 4. See In re Anchor Assurance Co. (1870), L.R. 5 Ch. App. 632, at p. 638. 5. See Ivamy, General Principles of Insurance Law, 5th ed. (1986), pp. 249-250.

In this case, there is no basis upon which to reach any conclusion other than that each renewal of the Blanket Fidelity Policy required the consent of both the respondent and the appellant underwriters. Accordingly, each renewal when it occurred constituted a new contract.

However, the fact that a fresh contract came into existence upon each renewal does not necessarily mean that the term "Policy Period", as used in the Blanket Fidelity Policy, was not capable of meaning the aggregate of the periods during which each insurance contract was in force. The question is, again, ultimately one of construction. The term "policy" is popularly used to refer to the terms and conditions of insurance which continue over a number of years, notwithstanding that renewal of the policy results in a fresh contract from year to year [18] . Thus, it has been recognized that a reference to a "policy" may be a reference to "a person's condition or state of being insured" [19] , notwithstanding that a contract or series of contracts is necessary to give the policy legal force.

1. See Stokell v Heywood , [1897] 1 Ch., at p. 464; Davies v Sprod , [1943] S.A.S.R. 53, at p. 57. 2. See Re Kerr, [1943] S.A.S.R., at p. 15.

Nevertheless, we do not think that the term "Policy Period" in the Blanket Fidelity Policy in this case can be construed as referring to the aggregate of the initial period of insurance and each of the periods for which the insurance contract was renewed. The loss covered under the policy was limited to loss incurred during the "Policy Period" and discovered not later than one year from the end of the "Policy Period". That period was defined in Item 2 of the Declarations as "12 months from noon on 30th April, 1985". It was not, for example, defined as meaning "12 months from noon on 30th April, 1985 and any subsequent period for which the Insured shall pay or agree to pay and the Insurer shall accept or agree to accept a premium for the renewal of this Policy".

The insurance and renewal certificates afford little assistance. The certificate for the year ending 30 April 1986 spoke of insurance "from loss or damage on the terms and conditions of the [Blanket Fidelity] Policy [f]rom 30 April 1985 [t]o 30 April 1986 and for such further period or periods as may be mutually agreed upon". Subsequent renewal notices stated that "This insurance is renewed for a further period of 12 months subject to the terms and conditions of the policy". In these later notices, the sum insured was expressed to be a lump sum "any one employee/all employees any one period of insurance" (or similar words). The reference to "any one period of insurance" is, however, suggestive of successive periods of insurance which are coextensive with the periods of renewal.

The respondent places some reliance on Section 9 of the Conditions and Limitations. That section provided:

Regardless of the number of years this Policy shall continue in force and the number of premiums which shall be payable or paid, the limit of liability stated in Item 3 of the Declarations shall not be cumulative from year to year or period to period. This provision clearly contemplates that the Blanket Fidelity Policy might remain in effect for a number of years, but it is not inconsistent with each contract of insurance constituted by renewal being coextensive with the Policy Period. The effect of the provision is that Item 3 of the Declarations applied to the whole of the "Policy Period"; to say that it indicates that the Policy Period encompassed the aggregate of the periods of renewal simply begs the question. Similarly, other references in the Blanket Fidelity Policy to periods within the Policy Period - for example, the reference in General Agreement A to "the current premium period" - do not necessarily indicate that the aggregate of the periods of renewal were included within the Policy Period.

The respondent further contends that the reference in General Agreement C to "any prior policy of insurance" would, upon the interpretation advanced by the appellant underwriters, include any prior contract of insurance and would therefore include prior renewals of the Blanket Fidelity Policy, a result which is said to render General Agreement C otherwise largely superfluous. But the phrase "any prior policy of insurance" is followed by the words "which prior policy is terminated, cancelled or allowed to expire as of the time of [the] substitution " of the coverage of the Blanket Fidelity Policy for that prior policy of insurance. Those words are hardly apt to describe the situation where the terms and conditions of a policy (except for the amount insured and the premium payable) remain the same with successive renewals. In our view, General Agreement C was intended to apply to the substitution of the Blanket Fidelity Policy for a different policy and not to the renewal of the same policy. The Industrial Special Risks Policy was such a different policy and so was within the contemplation of General Agreement C.

The respondent also points to the fact that the same policy number was retained despite successive renewals of the Blanket Fidelity Policy and that the renewal certificates did not purport to replace any previous policy but were simply expressed to be subject to the terms and conditions of "the policy". But the retention of the same number for the Blanket Fidelity Policy throughout the periods of renewal is inconclusive and the reference to the terms and conditions of "the policy" in the renewals involves the use of the word "policy" in the sense of a document that contains the terms and conditions of an insurance contract.

As is apparent from what we have said we do not find the Blanket Fidelity Policy "obscurely worded" [20] in any relevant sense and consequently there is no scope for the application of the contra proferentem rule on which the respondent relies.

1. See Western Australian Bank v Royal Insurance Co. (1908), 5 C.L.R. 533, at p. 567, per O'Connor J.

For these reasons it is our view that the term "Policy Period" as used in Section 1 of the Conditions and Limitations of the Blanket Fidelity Policy means the annual period for which the Blanket Fidelity Policy was initially taken out and thereafter the successive periods of renewal of one year each. Accordingly, we would allow the appeal.

The respondent, Edwards Dunlop & Co. Ltd. ("Edwards Dunlop"), was insured between 1983 and 1988 with the appellant, C.E. Heath Underwriting & Insurance (Australia) Pty Ltd ("the insurer"), against fraud on the part of its employees. In July or August 1988, it was discovered that Edwards Dunlop had been systematically defrauded by one of its employees over a period extending back to 1982. The question in this appeal is the extent to which the losses involved can be recovered by way of insurance.

The facts and the terms of the insurance policies relevant to this appeal are set out in the joint judgment of Dawson, Toohey and McHugh JJ. I shall repeat them only to the extent necessary to make clear my reasons for concluding that Edwards Dunlop is entitled to recover those, but only those, losses it sustained as a result of fraudulent acts committed between 30 April 1986 and their discovery in 1988.

There were two distinct periods of insurance: the first was from 31 May 1983 until 30 April 1985 when Edwards Dunlop was insured under an Industrial Special Risks Policy ("the I.S.R. Policy"); the second was from 30 April 1985 when it took out a Blanket Fidelity Policy ("the Fidelity Policy"). The Fidelity Policy was renewed on 30 April in each of the following years, the last relevant renewal occurring on 30 April 1988 for a further period of one year.

It is common ground that losses sustained prior to 31 May 1983 when Edwards Dunlop took out the I.S.R. Policy are not recoverable, being covered neither by that policy nor by the Fidelity Policy. It is also common ground that losses are recoverable only under the Fidelity Policy: the I.S.R. Policy, which terminated when the Fidelity Policy was taken out on 30 April 1985, expressly provided that the insurer's liability under that policy should not "include any loss unless discovered within twelve months of [its] termination".

Section 1 of the Fidelity Policy relevantly provides:

Loss is covered under this Policy only if discovered not later than one year from the end of the Policy Period. Subject to General Agreement C, this Policy applies only to loss sustained by the Insured through fraudulent or dishonest acts committed during the Policy Period.

If the coverage of this Policy is substituted for any prior policy of insurance which prior policy is terminated, cancelled or allowed to expire as of the time of such substitution, this Policy applies to loss which is discovered as provided in Section 1 and which would have been recoverable under such prior policy except for the fact that the time within which to discover loss thereunder had expired. It is not disputed that, on 30 April 1985, the Fidelity Policy was substituted for the I.S.R. Policy and that, subject to discovery as required by that policy, it covered losses resulting from fraudulent acts committed between 31 May 1983 and 30 April 1985.

It is apparent from the terms of Section 1 and those of General Agreement C that, as a matter of strict logic, there are two exercises to be undertaken before it can be said whether losses may be recovered under the Fidelity Policy. The first is to identify the loss insured against; the second is to identify the period in which it must be discovered, that being the period signified in Section 1 by the words "not later than one year from the end of the Policy Period".

The argument for Edwards Dunlop was that "Policy Period" means the period commencing when the Fidelity Policy was first issued and running for so long as the policy was or is renewed. On that approach, all losses resulting from fraudulent acts committed between April 1985 and their discovery in 1988 are covered under Section 1 of the Fidelity Policy as "loss[es] sustained through acts committed during the Policy Period". And, losses resulting from acts committed during the cover of the I.S.R. Policy are recoverable pursuant to General Agreement C as losses which "would have been recoverable under [a] prior policy except for the fact that the time within which to discover loss ha[s] expired", they being losses discovered as provided in Section 1 because they were discovered during the Policy Period. On that approach it is largely unnecessary to engage in the separate exercises involved in ascertaining what losses are insured against by the Fidelity Policy and the period in which they are to be discovered.

The argument on behalf of the insurer was that the Fidelity Policy, when first issued, was a contract for one year and that, each time it was renewed, a new contract came into existence for the following year. In that context it was then put that, cancellation apart, "Policy Period" in Section 1 refers to a particular period of one year, being the year involved in each contract of insurance. That approach, if correct, enables it to be said that the losses insured against in each contract are recoverable only if discovered during the year following that particular contract. But it also requires an examination of Section 1 and General Condition C to ascertain precisely what losses are insured against and are, thus, recoverable if discovered within that period.

So far as the meaning of "Policy Period" is concerned, I adopt what is said in the joint judgment of Dawson, Toohey and McHugh JJ. In particular, I agree with their Honours' conclusions that each renewal of the Fidelity Policy constituted a new contract of insurance and that, cancellation aside, the expression "Policy Period" in Section 1 means the particular period of one year involved in each contract and not the aggregate of those periods. But, as already indicated, that does not resolve the question raised by this appeal.

The second part of Section 1 identifies the losses insured against in each contract of insurance as those which "[s]ubject to General Agreement C [are] sustained through fraudulent or dishonest acts committed during the Policy Period". It is quite clear from the terms of that agreement that the words "[s]ubject to General Agreement C" are words of extension and not words of limitation. Their effect is to extend the losses insured against to include not only those losses "sustained through fraudulent acts committed during the Policy Period" but also those which, but for the expiry of a discovery period, would have been recoverable under a prior policy for which the current policy was substituted, provided they are discovered as required in Section 1, namely, "not later than one year from the end of the [new] Policy Period."

Two interrelated questions arise with respect to General Agreement C. The first is whether the words "substituted" and "substitution" comprehend the entering into a fresh contract of insurance, whether by renewal or otherwise, upon the expiry of the cover provided by an earlier policy. And if "substituted" does have that meaning, a question then arises whether General Agreement C is concerned with renewal of the Fidelity Policy or only with the substitution of that policy for some quite different policy. More precisely, the question is whether, in terms of General Condition C, on renewal of the Fidelity Policy, the coverage of "[that] policy is substituted for [a] prior policy which is terminated, cancelled or allowed to expire as at the time of such substitution".

The word "substitute" ordinarily signifies the putting of one thing in the place of another, with the former performing the function properly or normally performed by the latter. However, it can also signify replacement, in the sense of taking the place of something that no longer performs or is no longer able to perform its ordinary function. In a context involving insurance policies which have been "terminated, cancelled or allowed to expire" and which, at least ordinarily, are contracts for a finite period, the notion of substituted coverage extends, in my view, to embrace insurance cover effected by a policy which replaces or takes over from an earlier policy upon the termination or expiry of its cover and the words "substituted" and "substitution" in General Agreement C should be so read.

Once it is accepted that each renewal of the Fidelity Policy effects a new contract of insurance, the words "this Policy" in General Agreement C must be taken to refer to the Fidelity Policy as incorporated in the contract current from time to time. And when so read, the words "prior policy" inevitably include the Fidelity Policy as incorporated in the contract for the previous year. On that basis, General Agreement C must be read as referring to the situation involved whenever one contract of insurance takes over from another, including the situation brought about by renewal of the Fidelity Policy. And where contracts are made with effect from the day on which the previous contract comes to an end, as they were in this case, the prior policy is one that "is terminated or allowed to expire as of the time" the new contract is made, or, to use the language of General Agreement C, as of the time the new policy is "substituted" for it.

The consequence of the words "[s]ubject to General Agreement C" in Section 1 of the Fidelity Policy is that each contract of insurance effected by a renewal of that policy was a contract insuring against loss by fraud which was discovered "not later than one year from the end of [that] Policy Period" and which, but for the expiry of a discovery period, would have been recoverable under the contract effected by the previous renewal as well as against loss resulting from fraud committed during the year of that contract.

It was argued on behalf of Edwards Dunlop that, if General Agreement C operates on renewal of the Fidelity Policy, it operates so that the contract of insurance effected by renewal covers loss resulting from fraud committed at any time while the Fidelity Policy was on foot and at any time while the I.S.R. Policy was on foot. So far as the I.S.R. Policy is concerned, this is because that policy was one for which the Fidelity Policy was substituted and, subject to its discovery period, one that covered all losses sustained as the result of fraudulent acts committed between 31 May 1983 and 30 April 1985.

The argument that General Agreement C operates so as to render recoverable all losses sustained as the result of fraudulent acts committed after 31 May 1983 proceeds on the basis that the loss insured against in a contract of insurance effected by a renewal of the Fidelity Policy includes that recoverable under a prior policy which, in turn, covers that recoverable under the previous policy and so on indefinitely, provided that there was a substitution of one policy for another as contemplated in that agreement. That overlooks the fact that, upon renewal, the new contract insures against prior undiscovered loss only if discovered as provided in Section 1, namely, "not later than one year from the end of the [new] Policy Period". Unless discovered within that period, the loss is simply not a loss insured against in the new contract. Not being insured against, it is not a loss that is recoverable under that new policy, much less one that, given the passage of sufficient time, "would have been recoverable" but for the expiry of a discovery period.

The language of General Agreement C may leave something to be desired, but, in general terms, it effects what one would expect of insurance of this kind, namely, cover for a rolling period so that, on each renewal, cover begins to roll forward for the following year but falls away for some past period unless loss referable to that period has already been discovered. Its precise effect varies according to the point in the Policy Period when loss is discovered: there is longer cover if discovery is late in the life of the policy period and shorter cover if earlier in that period.

In the present case the losses were discovered in July or August 1988 and, thus, were discovered during the cover of the 1988 policy (12 months from 30 April 1988) and during the discovery period of the 1987 policy, being "not later than one year from the end of the [1987] Policy Period", that "Policy Period" having ended on 30 April 1988.

The 1987 policy insured against losses sustained as the result of acts committed during its cover and losses which "would have been recoverable" as losses "sustained through fraudulent acts committed during [the 1986] Policy Period" but for the discovery period in the 1986 policy, provided they were discovered, as they were, within the discovery period of the 1987 policy. As already explained, the 1987 policy did not insure against loss sustained through fraudulent acts committed in 1985 or during any earlier period as no part of that loss was insured against unless discovered, at the very latest, immediately prior to the expiry of the discovery period for the 1986 policy.

The 1988 policy insured against losses sustained as the result of acts committed during the period of its cover, namely, on and after 30 April 1988. So far as is presently relevant, they were the only losses insured against by that policy: it simply did not apply to losses sustained during the cover of the 1987 policy, they being losses which, having been discovered within the discovery period of that policy, were recoverable under it and, hence, not losses that "would have been recoverable" under it but for the fact that its discovery period had expired.

The judgment at first instance was to the effect that Edwards Dunlop could recover only those losses sustained as a result of fraudulent acts committed after 30 April 1987. That failed to give effect to General Agreement C under which losses sustained as the result of acts committed during the cover of the 1986 policy were, in the circumstances, insured against by the 1987 policy and recoverable under it. On the other hand, the Court of Appeal of the Supreme Court of New South Wales held that Edwards Dunlop was entitled to recover for all losses sustained through fraudulent acts committed after 31 May 1983. The appeal must be allowed, but it should only be allowed to the extent that the decision and order of the Court of Appeal allow for the recovery of losses sustained through fraudulent acts committed before 30 April 1986.