Making some crude calculations and assuming that each person is a cylinder, then ignoring the losses from the ends, heat loss is proportional to the circumference of each circle (cylinder cross section) divided by the area of the circle. So the heat loss coefficient for each is 2/radius.
Assuming that each cylinder huddles, one next to the other in line, then, crudely speaking, the heat loss coefficient for the line is 1/radius.
Summarising, these crude and incomplete calculations suggest that two people huddled together may have something approaching 1/4 of the heat loss from their surface area than two people insulated separately (2 times 2/r).
I don't actually believe these calculations, because of the assumptions made and the other types of heat loss ignored, but they do indicate why mitts keep your fingers warmer than gloves and why a group shelter or Blizzard Tube is exceptionally effective.
...And some credence may be given to the concept if you have experienced how much warmer it is in the middle of a large crowd on a cold day.