As you'll see in eq. 3 of this paper

[link|http://qoptics.physics.harvard.edu/RMP75_457.pdf|http://qoptics.physi...edu/RMP75_457.pdf]

the reference velocity is the "group velocity". To have a group velocity, you need a group.

The idea is that one considers the propagation of photons in an already existing photon "bath". That provides the group needed to make a group velocity - so even when they speak of "single photons", they are thought of in reference to the surrounding bath.

It should also be remembered that a single, bare photon cannot be localized at all - it has no rest frame and it is not even possible to assign coordinates to it - in fact the wave function of a single photon in a pefect vacuum extends to inifinity, so in a sense it is "everywhere" at once. All one can actually do with a single photon is say when it has interacted with something. (The position is undefined because for any photon, the momentum is exactly known, by definition - then apply the Heisenberg principle etc. Even in classical electrodynamics, the idea of an individual light "BB" moving at the speed of light is untenable.) For this reason (and others), it is not even legitimate to speak of individual photons in a complex system - there is simply the electromagnetic field as a whole. To give an example - when one passes light through a colored filter in a perfect vacuum, it is tempting to think of the transmitted light as just the original light minus certain frequencies - but in truth the original light dies at soon as it strikes the surface of the filter - when it dies, it sets up excitations in the atoms of the first few layers of the filter, which in turn set up exicitations in deeper layers etc. etc. until finally after countless emission and absorption events, this disturbance has propagated all the way across the filter. In fact the photons of light that allow you to read this print, are born right inside the rods and cones of your retina, not propagated across the space between the screen and your eye.

The speed represented by a group velocity is not phyiscal in the same sense as the speed of bullet. The best analogy is the intersection point of a pair of perfect scissors. If one blade of the scissors is represented by the x-axis, and the other by a line through the point (0,-1) with positive slope, then as the scissors close, the point of intersection races off to infinity. The intersection point, of course, carries no energy or momentum and is purely conventional. The relativistic speed of light applies to physical objects that carry mass. The key point is that group velocities cannot be associated with transfer of energy from place to place.

These photon traps are similar in principle to the filters and traps used in electronic music to induce signal delays (for reverb effects, etc.)

One more point, what one actually measures is this type of thing is occupation number - one doesn't measure the "speed" in the same sense as one does that of a runner, with a stopwatch.

This research is of course very significant for optical memory devices, but it's a shame that the press and sometimes even the researchers miss basic points of interpretation.