THE MOST basic function of any state is to maintain order among its inhabitants and to defend them from outside attack. To do this, most states have made it their business, sometimes their only business, to recruit and maintain a military force. In pre-modern states, the payment of the military and the collection of taxes to do this was usually the most important administrative function of the government, to the extent that the whole structure of the state was determined by the way in which this was done.

There were a variety of possible models, ranging from the highly centralised to the extremely dispersed. At one end of the spectrum were states like the ‘Abbāsid Caliphate under al-Mu'ta.sim, where taxes were collected from all over the empire and brought to the capital where they were used to pay an elite, professional military force. At the opposite end are “states” like France in the tenth century, when virtually all vestiges of central government had disappeared and the military subsisted by taking over areas of land and exploiting them to maintain themselves without reference to any central authority. Most states used a mixture of these two in varying combinations; such was the later Roman Empire with its combination of field armies and limitanei.

The way in which the military is paid affects fundamentally the whole structure of the state. To put it at its simplest, a centralised system will put power into the hands of the central authorities who collect and distribute military pay, a decentralised system which assigns this power to local authorities , shifts the balance of power into the hands of provincial administrators. A system in which the military secure their own resources from land under their direct control makes central administration virtually redundant. The system of payment of the army determines the structure of power and the nature and scope of the “civil” aḍministration.

All these points are familiar territory, especially for western European mediaevalists, brought up on discussions of the nature of power within a feudal system. Much less has been written about these questions in connection with the early Islamic state and this paper is an attempt to suggest answers to some of them in the Marwānid period (684-750). The Marwānids ruled, in one sense or another, over a vast empire that stretched from the Atlantic to the Indus, but while the governors in all those areas were in theory appointed by and responsible to the caliph in Damascus, it is much less clear how much effective control the Caliphs exercised in the various provinces of their far-flung domains. The question of the fiscal structure of the Caliphate is crucial to understanding this. If the Marwanid Caliphs were able to collect and exploit the tax revenues of the provinces at will, then the state was clearly both powerful and centralised; if, on the other hand, the revenues collected in the provinces were spent there without corning to a central treasury and according to the demands of the local Muslims, then the state was a loose federation and the role of the central government very restricted.