Firstly, if you stipulate for slave A and there are two slaves called A, which slave the stipulation is binding for depends on evidence extraneous to the verbal contract. For example, if a hand is placed on one slave (D.34.5.21.pr.)
Secondly, in the situation where a document has been used as evidence of a contract but the document is ambiguous (which is where agreement is most important) the texts discuss two possible responses. This ambiguity may be resolved against the stipulator (the approach taken by Celsus and Ulpian) or the mistake may make the contract void (the approach taken by Venellius and Paul and followed in Justinian’s Institutes.)
The Greeks, however, used purely written contracts, which clearly had some influence on Roman law, as writing was used. However, Riccobonno has argued that the oral contract was sufficient and the documents merely supplemented the written contract as evidence. In C.4.32.1 (circa 160AD) there is a question which was sent to the emperor involving a situation where a document witnesses a promise to pay back a loan but does not mention the requirement of interest, which was a part of the oral promise. The response is that the interest is still enforceable. This fits with D.50.17.92 which says a stipulation which is recorded wrongly is still valid. Also, D.22.214.171.124 is an example of a written stipulation, and it speaks in the past tense, that T had asked and M had promised, which again suggests that the written documents performed and evidential function.
There is some dispute about whether or not the stipulatio degenerated and became a written contract. Nicholas has argued that the written document became very strong evidence of an oral agreement (circa. 140AD – D.126.96.36.199.) There was thus a presumption that an oral promise had been made, unless it could be proven that the parties had not met. By 201AD this is stated explicitly in the texts, for example, in S.7.2. This approach demonstrates an attempt by the Roman lawyers to reconcile Greek contracts, which purported to be solely written, with the Roman world, which required an oral exchange.
Diosidi has argued that the stipulatio degenerated further, so far that although theoretically you must have an oral exchange, in practice this was not necessary. The first text he uses to support his argument is G.3.1.34 (circa 130AD.) However, this text refers to chinographs and synographs, (non Roman terms) and it seems a better interpretation of it is that it is simply describing the situation in Greece. He also refers to D.45.1.30 (circa. 220 AD) in which a written document purports to be a contract, however, this is not incompatible with the arguments of Nicholas.
Nicholas has argued that veluti here means that is to say and that the stipulation started off as being a set list of words that could be used. There are examples in the texts of other words being used but he argues that most of these are interpolations, for example D.19.2.54.pr (which is generally accepted as being interpolated) and D.188.8.131.52 which appears to have been torn from its original context as it is inelegant and formed in bad grammar. He accepts that some words may have been added to the list (such as Inst 3.1.) and argues that Justinian was attempting to revive classical law but at the same time, reconcile written and oral practice, so he interprets this text to mean that the list becomes an unbounded list at this point.
Other academics have contested this. Watson argues that since you can destroy a stipulation without formal words, you can probably create one in the same fashion. Winkler says that Gaius always uses veluti to mean for example. Van Oven points to the fact that G.3.92 does not say that the words create a stipulation but that they create a verbal obligation. He has argued that given that there are another two types of verbal obligations not in the list, the list in inconclusive. Also, it seems strange that a limited list should contain such informal words as dari, meaning simply to give (of the 6 examples of words given, the first four are very formal, but the others are very common.) This seems, at first glance, to be undesirable, as parties would be bound far too easily. However, Ibbetson has pointed to the requirement of intention and said that this means that an unbounded list was not a problem. Taking this approach, Riccobono thus argues that C.8.37.10 means that there was no longer any need for question and answer.
D.45.1.2 suggests that the degeneration may have gone further still, requiring no correspondence between question and answer. This text says that if one party saus “Will you give?” and the other says “why not?” he is bound, but if he merely nods, there is no civil or natural obligation. The example of a guarantor is then given. The text is clearly abbreviated, and it seems unlikely that the stipulation did ever degenerate this far as if this was the case, we would expect the many references to agreement between question and answer to have been excised from the other texts.
D.45.1.16 says that if the question is in Latin and the reply is in Greek, the obligation is settled providing the two correspond. This seems to suggest that there was not a closed list of words. Importantly, it then goes on to say that other languages can only produce an obligation, not an action. However, the examples given are all of dead languages – which may be significant (it may be that other languages could have been used.) An obligation is only produced where both parties understand each other, but this may be through an interpreter. The last clause suggests that the text may have been interpolated, as the use of an interpreter seems unlikely as it would be highly problematic because the interpretator might lie as to what was being said.
That the text expressly rebuts the idea that a reply may be given on the same day after a party has undertaken some other business, suggests that such things were valid at some point in Roman law, although it could be an interpolation. D.184.108.40.206 says that a person answering on the same day is bound but someone answering on the next day is not. An analogy may be drawn with a legal trial, where things had to take place in one day but they would divide the day, so it was technically the same day, even though not chronologically. The emphasis placed was on factual continuity and legal bondedness.
Stipulation for when I die: This is valid and is taken to mean I will perform when I am dying – the performance is due in the last moments of life (G.3.100). However, a stipulation for the day before death is not valid (Inst 3.19.15)
Stipulation for after death: This was probably not permissible during classical law because the obligation resides solely in a third party (the heir) – G.3.100. However, Ulpian, when he states the rule that one party cannot promise for another, makes an exception for heirs, which may mean that it was permissible. By the time of Justinian, the clear inconsistency between permitting a stipulation for when I die but not for after death or for the day before death, was removed, as both were now valid – Inst.3.19.13.
Stipulation for 10, answer given is 20: According to G.3.102 this would be void. However, according to D.220.127.116.11 it would be valid for 10. If the stipulation is for slave a and slave b and the answer is slave a, it will be valid for that one slave (D.18.104.22.168.) Ulpian argued there are as many stipulations as things stipulated for. This may seem unfair, but it is up to the promisee whether or not he enforces the promise.
Stipulation for a or b, where the answer given is a: This is void in the case of slaves (D.22.214.171.124) but valid in the case of money (D.126.96.36.199.) The difference arises because money can’t be destroyed in the way that slaves can be, so a stipulation for slave a or b is fundamentally different from one for a because in the first instance if a dies, the stipulation would be valid for b, whereas in the second, there would be no contract.
A resolutive condition is where there is an existing debt which is destroyed on the occurrence of an event. For example, I will give you £10 unless you remarry. This raises problems in this instances as the only time when you can be certain that you have not remarried is on your death, at which point it is of no use.
A suspensive condition is one where there is no duty to pay unless the event happens. This was very useful. It could be used to create a penalty clause, concentrating the mind on performance because if the building was not built you would have to pay money. It also allowed for the determination of damages on such an event, rather than leaving it in the hands of the iudex. Leaving it in the hands of the iudex was uncertain because of problems of evidence, the non availability of specific performance and that the damages were discretionary and thus uncertain (Inst.3.15.7,) it was also faster to do it this way. These conditions could also be used to prevent appeal against the decision of an arbiter by putting a penalty in place if his decision was not kept to. It could also be used to ensure payment to a third party by making a condition that if a sum was not paid to T, the money would have to be paid to me (which could not be achieved directly due to the rules relating to payment of third parties.)
There were, however, a few types of conditions which were problematic:
Immoral conditions – these were void outright e.g. promise money to someone if they don’t marry someone (C.3.82)
Impossible conditions e.g. a cow with no legs jumping - These were void, and if found in a will, the section which was impossible would be struck out (G.3.98.)
Conditions which are always fulfilled – If £10 is promised if S doesn’t win the lottery and S can never win the lottery because he is a slave, the money is always due (D.44.1.7 – repeated by Justinian.)
Risk bearing in contracts – If A promises £10 if B doesn’t give £5 to S, and S dies the risk falls on A. However, it can be redrafted as £10 will be given is £5 is not given to S. If S then dies the risk falls on ????
Perplexing conditions e.g. in wills - that S is to become free and when he is free, he is to be heir (D.188.8.131.52) Romans normally saved them. For example, in this case they regarded the man as both free and bound at the same time.
Preposterous conditions – Obligation due before you can know if it is true. J.3.9.14 says that it used to be that preposterous conditions were invalid but Justinian changed the law. However, it doesn’t seem to be that it was invalid by the time of Paul and Africanus this may have been an interpolated, or the lawyers may have reached this point by later classical law. However, contrasting the examples given in these two texts suggests that there may have been a fundamental different between the two types of preposterous conditions. In J.3.9.14 the example given if promising to pay on Wednesday if the ship arrives on Friday. In D.45.1.126.pr it is that the party will pay 10 per year from today if S becomes counsel. In the second example, it could be that the obligation arises when S becomes counsel, however, backdated payments will have to be paid from this point now. The Code 6.3.52 also speaks on this matter, however it argues that they are invalid and valid, and appears to have been something like lecture notes, updated with the changes made by Justinian but without removing the old stuff.
Actio ex stipulatu – This is a claim for an uncertain thing and would give unliquidated damages e.g. failure to do something. The procedure was that you would quote the words of the stipulatio and the iudex would determine the amount due. This meant that there was leeway in determining the amount of damages owed (so that the stipulation was theoretically a stricti iuris contract was not a problem) and was also easier as you did not have to say how much you wanted to claim, and risk asking for the wrong amount.
Promise to give: stipulatio dari Promises to do: actio ex stipulatu
Duress - Metus was introduced as a defence at a later stage. The exact meaning of duress is uncelear from the texts, however, it seems that it must have been severe and contrary to sound morals (D.184.108.40.206) affecting a man of resolute character (D.4.2.6) and probably resulting in a fear of serious evil (Ulpian quoting Labeo – D.4.2.5.) That the requirements were so strict is to be expected – if the requirements were not so strict you might have parties arguing, for example, that they had been forced to enter a loan because they were poor (economic duress.) Dolus was included in metus, but since metus was enforceable against a third party whilst dolus was not, metus was a better defence to use (D.220.127.116.11.)