ABSTRACT
Many attempts to measure the speed of light clearly involve the measurement of light travelling to a given point and back; so the speed measured is that for the ‘round-trip’. For example, Fizeau’s method, a standard experiment in the mid-nineteenth century, involves light travelling from a source to a distant mirror and back to an observer close to the original source; again, Foucault’s method, which replaced Fizeau’s approach as the standard procedure, also measures the speed of light travelling around a closed path. 1 Numerous actual and thought experiments have been suggested in order to help us measure the one-way speed of light. Wesley Salmon argues that all such attempts involve a fatal flaw: any experiment designed to measure the one-way speed actually involves either a round-trip measurement or some other problematic assumption. 2 Hence, we are told, the claim that the speed of light (in vacua) in a given direction is always c is essentially a non-trivial convention. This has an immediate consequence for our definition of simultaneity. The Special Theory of Relativity (STR) demands a change in the way we think about simultaneity. The idea of a fixed ‘invariant’ speed of light is a key element in STR’s account of simultaneity. The light cone, which represents light spreading out in all directions from an event, provides the foundation for all judgements of simultaneity. 3 We can no longer rely on some vast cosmic time-slice providing a plane of simultaneity for all observers. We use light rays to synchronise events and the judgement that two events are simultaneous depends on how light rays propagate between them. If we want to be sure that a nearby sequence of events synchronises with a distant sequence in our frame of reference, then we typically bounce light rays back and forth to check that the sequences match. The shuttling light rays transmit the latest information about each of the sequences. If the round-trip takes 10 seconds, then we would generally feel confident in the assertion that the latest information received always relates to an event 5 seconds old. But we need to be sure that the time the light ray takes to travel from us to the distant sequence is the same as the time for the return trip. Otherwise our judgement of simultaneity between pairs of events in the sequences will be impaired; see Figure 9 above. Therefore, if our assertion that the light travels at the same speed in both directions is conventional, then our judgements of synchrony and simultaneity will also have conventional characteristics. 4
