Electronic computers have been available commercially for about 20 years. Actuarial interest was soon shown, but although a number of papers have appeared over the years there has been little recent discussion of how computers can best serve the profession.

Many actuaries have little computing experience. To them it may appear a daunting task to learn to use a computer, because many instruction manuals seem, at first sight, extensive and complex. Our principal aim is to demonstrate to them that given suitable permanent facilities it is quite easy to use a computer for actuarial calculations, however complex these may be. We suggest that the complexity of most existing actuarial programming systems is unnecessary, and may have impeded the understanding of the computer's potential by actuaries.

A time-honoured convention in accounting has been that accounts should be based on the principle of historic cost, namely that all items should be recorded in terms of the purchasing power of the pound at the date of each transaction. This convention has the virtue that the accounts are based largely on factual monetary transactions and fewer items need be determined subjectively. It is a valid convention so long as the value of money remains constant, but in a period of inflation, accounts drawn up on this basis become distorted, and the higher the rate of inflation the greater the distortion. For example, amounts based on historic cost which are set aside for depreciation of plant and machinery will, in a period of rapid inflation, be totally inadequate either to provide funds for the eventual replacement of those assets or to maintain the real value of the shareholders' original capital investment. Similarly profits are overstated by the inclusion of profits on stock which arise solely from a general increase in price levels. Again, no account is taken of the real cost of holding cash or other monetary assets when money is losing its purchasing power. Conversely, no credit is taken for the gain derived from having borrowed money, when the liability for repayment of the loan is in real terms reduced.

The results of the investigation from 1945 to 1959 were presented in a paper to the Institute by R. E. Hayward and B. C. Lucena (J.I.A. 91, 286) and readers are referred to this for further detail on practical aspects of the investigation. This further report owes its existence to all those originally involved as well as those specifically involved in the follow-up. Professor J. M. Malins and Dr M. G. FitzGerald have continued to give us the help needed to obtain the required data from the hospital records. Those of our Society particularly involved in producing this follow-up report include B. C. Lucena, R. Betteridge, A. Roberts and M. Willett.

Some 20 to 30 years ago considerable progress was made in understanding the variance of mortality rates and, for assured lives data, the effect on it of the presence of duplicate policies, e.g. Redington and Michaelson (1940), Seal (1941, 1947, 1948 and 1949), Daw (1945 and 1951), Vajda (1945), Solomon (1948), Beard (1951), Jager (1953) and Continuous Mortality Investigation (C.M.I.) (1957). I have the impression that since that time this understanding has tended to regress. For example it is disappointing to find that the whole course of reading on the analysis of mortality data makes practically no mention of the effect of duplicate policies in increasing the variance of mortality rates. Yet this is a matter which can have a substantial effect on the results of testing a mortality table graduation as is recognized in C.M.I. (1974). It is salutory to find that G. H. Hardy (1882) was aware that the presence of duplicates increased the variance although he did not express it in quite that way.