The key to this is the distinction between the dispositive and the evidentiary use of writing.

A dispositive use of writing occurs where the law specifies the written word as the means by which a legal effect is achieved, here the making of a contract. The effect is triggered by the document as such, precisely because of the writing.

Evidentiary use of writing occurs where a document is drawn up not for its own sake but to prove that some other dispositive event has happened.

Take the case of the conveyance called mancipation. The law specifies that dominium over Italic land cannot be passed from one person to another except through this ceremony with bronze, scales and witnesses. Suppose, however, that in order to avoid a journey we just write out that the mancipation of the land took place on such and such a day before such and such witnesses. As alienee you may be perfectly happy with that. If the document, duly authenticated, says that the mancipation happened, it will be difficult for anyone to say it did not. A layman may say that the document itself has the effect of transferring dominium. But a lawyer will say that that analysis is obvi­ously wrong. The dispositive event is the mancipation. The document obtains its utility because it is very good evidence of the mancipation having happened. Even if the mancipation had happened the docu­mentary record would probably have been made. That is only sensible. You always need to be sure of having good and convenient evidence.

If the document seems to work on its own, that is only because people are exploiting the practical realities of the relationship between written evi­dence and the dispositive ceremony. The document itself is only evidence.

This example is taken from outside the field of contract, mancipa­tion being a conveyance rather than a contract.

The category of contracts litteris should not have anything to do with the merely evidential use of writing. Just as contracts verbis are those made by word of mouth and contracts consensu are those made by agreement however reached, so contracts litteris should be contracts made by writing, those for which writing is the dispositive event.

It is almost true that classical Roman law knew no such contract. That is, that its law of contract made no use at all of dispositive writing. Under the heading of contracts litteris Gaius puts forward only one tiny figure, already obsolescent if not obsolete. And his most important statements are negative. These negative statements are at G.3.134 and G.3.131-2. G.3.134 reads:

Praeterea, litterarum obligatio fieri videtur chirografis et syngrafis, id est, si quis debere se aut daturum se scribat, ita scilicet si eo nomine stipulatio non fiat. Quod genus obligationis proprium peregrinorum est.

Besides, an obligation based on writing seems to be made by chirographs and syngraphs. That is, where (supposing the case to be one in which no stipulation is made for the matter) one writes that he owes or will convey. But that type of obligation is peculiar to peregrines.

This is negative from the point of view of the Roman law.

In G.3.131—2 he is making a similar point about genuine cash-book entries (nomina arcaria). Suppose I take £10 from my cash-box (arca) and lend it to you. And I enter the loan in my cash-book. Does the writing put you under an obligation? No. There is no obligation unless you receive the money and if you do receive the money your obligation is â€?real’ (re). These entries should be said nullam facere obligationem sed obligationis factae testimonium praebere (to create no obligation but to evidence an obligation already created). And then he goes on to say that you cannot therefore maintain that peregrini are here bound by written entries (nomina) since they are bound only by the paying out of the loan, a contract indubitably iuris gentium. Again you detect a theme on these lines: can the Greek attitude to documents be explained within the Roman framework? If they regard themselves as bound by these nomina, they will be bound but not by the writing, only by the ius gentium contract re evidenced by the writing.

We are working backwards and so come last to the things which Gaius does present as true examples of contracts litteris. As well as genuine nomina recording loans, there are also what are called nomina transcripticia, transcriptive entries. The transaction involving these entries is also known as expensilatio, which might be translated �debit’ or �entering a debit’.

Gaius (G.3.128-30) says there are two types of transformation depending on whether the transcription is �from a matter to a person’ or �from a person to a person’. The former, a re in personam, happens where for instance you owe me the price of goods sold to you or hired to you. In my ledger I carry this to your debit, presumably only with your consent. What change is thus effected? I get a stricter form of action, for expensilatio gives rise to the old condictio. It is one of the traditional causes of strict ius civile debt. Sale and hire by contrast give rise to claims based on good faith, giving you more room to manoeuvre. Then the latter, a persona in personam, happens when, with the agreement of my debtor, I enter to your debit the sum which he owes me. And the effect of this implementation of an arrangement made between all three of us is to give me the condictio against you in place of him. In both these versions what seems to happen is that, without any money actually passing, a position is reached exactly as though the first obligation had been discharged by payment and the money had then been lent out again, either back to the payor (a re in personam) or to someone else, the new obligee (a persona in personam). I have spoken of the first obligation’s being �transformed’. The technical term for such transformations—extinctions and renewals—is �novation’.

There are several puzzles about expensilatio. One reason why we know so little is that it was at best obsolescent in Gaius’s time. No detailed discussion has survived. One question is whether the written entries really took their force from the paying out which they seem to simulate. Was the writing really effective only as evidence of an actual discharge and new loan? It is likely that that was the beginning of it. But probably the question was raised at some point, whether a defendant could resist liability by showing that the payments in and out had never in fact been made.

We have already seen that so far as concerns nomina arcaria, where the entry evidences an actual loan and there is no element of novation, Gaius scrupulously ascribes to the writing only evidential force: in his enim rei, non litterarum obligatio consistit; quippe non aliter valent quam si numerata sit pecunia (G.3.131): for in these the obligation arises re not litteris; since they are not valid except in the case in which the money has been paid out.

It looks as though, faced with a plaintiff relying on nomina (ledger entries), the defendant could escape by showing there had been no actual paying out to him, unless the plaintiff could show there had been a transcription of one of the two kinds. Within the scope of that �unless’ the writing had ceased to be merely evidential.

There is a vivid example of a transcriptio a re in personam in Cicero, De Officiis, 3.58-60. According to our way of naming cases this would be Pythius v. Canius, an action by a seller for the price of the thing sold, brought not on the sale itself but on the basis of expensilatio. Canius was looking for a holiday house near Syracuse. Pythius asked him to dinner at his own seaside villa. He persuaded the fishermen of the district to do their fishing on that day within view of the house, and he told Canius that the place was a centre of the fishing industry. Canius agreed to pay a fancy price.

The contract of sale rests on good faith, as we shall see. Pythius could not therefore have sued by action on that contract. His malafides would have given Canius a defence. But this deal had been transformed litteris as soon as made. Hence Pythius's claim lay by the strict condictio and not by the actio venditi. Canius had no defence, for the date was before the introduction of the exceptio doli by Aquilius Gallus. Pythius had, by his transcriptio a re (from the sale) in personam, acquired an action immune from the effects of his own dishonesty. The trick would not have worked once the defence of fraud had been introduced into the condictio. That praetorian innovation went far towards eliminating the difference between the stricti iuris and the bonae fidei actions.