The study of the structure of treaty texts often comes up against a stumbling block: these are large tablets, which have generally come down to us in a very incomplete form, either because the tablets have been destroyed deliberately, as in the case of Asarhaddon's Vassal Treaties at Nimrud, or because of a simple physical phenomenon: the larger the tablets, the less likely they are to remain intact, as shown by the sad state of most of the treaties discovered at Tell Leilan. The structure of the texts varies greatly from one period to another. There are too few dating from the 3rd millennium for a typology to be possible for this period.
In the Paleo-Babylonian period, texts were written unilaterally: each king committed himself to the wording of the oath sent to him by his counterpart, according to the procedure analyzed above. These texts, described as "great tablets", were drafted in three parts: a list of the gods by whom the oath was sworn, clauses and, finally, curses in the event of perjury. Differences in detail nevertheless emerge, clearly showing that there was no fixed pattern followed in all cases.
In the second half of the 2nd millennium, a number of innovations can be observed: the list of gods was moved after the clauses, and treaties often opened with a historical preamble. Among the Hittites, there was a clear difference between parity treaties and unequal treaties; the latter were written as a speech from the "suzerain" to his "vassal". By contrast, egalitarian treaties were now formulated bilaterally. The treaty between Hattusili III and Ramses II is a case in point, although the two known versions are not identical and not all clauses are reciprocal.
For the Neo-Assyrian period, scholars distinguish between loyalty oaths sworn by subjects or vassals and treaties concluded between kings; but the term adê used to designate these two categories emphasizes what they had in common, namely the oath that bound those who swore it.
