In English and in many other languages futurity is marked by the same element that, in different contexts, has a modal meaning. For instance, alongside the use of `will' in (1), in which `will' simply supplies a future time, there are uses of `will' as an epistemic modal, compatible with temporal reference to the present or to the past, as in (2) and (3).

(1)	I will present in her class tomorrow.
(2)	He will/must be stewing over his talk right now.
(3)	She will/must have left the island yesterday.

Can `will' be given a unified meaning, covering the cases of simple futurity as well as the uncontroversially modal uses? There are a number of positive suggestions to this question, often taking `will' to be a modal of prediction and its futurity a consequence of that reading. However, as desirable as a unified analysis may be or as good as the arguments against ambiguity are, there has been no fully convincing analysis of `will' as a modal in its simple future use that adequately addresses a number of prima facie arguments against it.

In this talk I will propose an analysis of `will' as a metaphysical necessity modal with a temporal interpretation identical to that of other modal auxiliaries. Specifically, modals are of the same type as aspectual operators, mapping from world-time properties to world-time properties. Their temporal perspective is fixed externally, by the operator whose scope they are directly under (present tense for `will', past tense or the perfect for `would'). If modals have the perfect in their immediate scope, as in (3), they exhibit a backward-shifting reading due to the effect of the perfect.

We still need to account for why `will' can be construed with a metaphysical modal base in (1) but not in (2) or (3). This contrast is part of a wider generalization: modals allow for a metaphysical reading only when the property they apply to is instantiated at a time in the future of their temporal perspective. I argue that this is due to the fact that in that case the (non)-instantiation of the property is not presupposed to be historically necessary. If we take common grounds to be unions of equivalence classes (historical alternatives) of worlds relative to the time of utterance, a metaphysical reading is excluded, in any context, when the modal takes scope directly under present tense and scopes over the perfect, as in (3), or when the modal takes scope directly under tense and combines with a predicate instantiated at a time including the time of utterance, as in (2). These are precisely the cases where the (non)-instantiation of the property the modal applies to is presupposed to be historically necessary relative to any common ground.