A new photo by Dan Berard shows what looks like an intermediate large redpoll, and should be added to the discussion in my previous post.
Dan says both of these birds were about the same size and both were larger than the other redpolls with them. This is interesting on its own because the paler bird looks larger in this photo. Assuming that Dan's observation is a more reliable measure of size, this shows the danger of putting too much emphasis on a single photo when judging size. [update 11 March - Dan points out that the pale bird is at the near edge of the railing, and the dark bird at the far edge, so that could account for the apparent size difference in the photo.]
The bird on the right is dark and heavily-streaked, approaching the "classic" rostrata, with heavy streaks on the breast and flanks. The bird on the left is distinctly paler and Hoary-like, with the pale gray ground color of the head and back, whitish flanks with fairly narrow streaks, etc. If the bird on the left was smaller, I would study it as a potential exilipes Hoary, but it has some odd dense fine streaking around the neck and breast, and the flank streaks are unusually broad for Hoary. It seems too pale to be rostrata, too dark to be hornemanni. Which is it?
Another note: This photo (with two large birds together) demonstrates the point that redpolls seem to travel in more or less cohesive groups within the big flocks - what we might call "subflocks". This is similar to the way geese sort out within a big flock, but the redpoll subgroups are not quite as cohesive. There have been quite a few reports this winter of small groups of redpolls with a high percentage of hoary-types (like 100%). And at Dan Berard's feeders the large birds seem to show up in waves - none in a group of fifty, and then 8 in a group of 30. What this means for identification is that the birds immediately adjacent to each other are more likely to be similar, and that comparing an "interesting" bird to those immediately around it might not be the best measure of how different the interesting bird is from the "typical". You should try to compare an interesting bird with other redpolls across the entire flock, and don't be surprised if studying one interesting bird reveals the presence of several more.
Update 11 March 2008: I wanted to add two more general cautions suggested by Dan Berard.
1) One of the things that makes size difficult to judge is the fluffiness of redpolls. Birds that are relaxed, with feathers puffed out and sitting more upright, can look larger than adjacent birds that are actively feeding. Be sure to consider what the bird is doing and watch a bird for several minutes if possible to get an accurate impression of size.
2) When redpolls fluff their feathers they reveal more white, making the dark flank streaks relatively less obvious. This is particularly obvious where the upper flank feathers overlap the folded wings, and these feathers can also be displaced and stick out over the wing just after a bird has landed. Dan says his eye is sometimes drawn to a flash of bright white, only to discover that it is not a pale redpoll but simply a redpoll with feathers fluffed or a few feathers askew.