Actually, neither value means a whole lot. Of course, extreme spread of 70 fps would probably show up as a wee bit larger group if your average velocity were sub 1000, but such an extreme spread in the 3000 fps range really don't much matter. Standard deviation has absolutely zero, no, none, relationship to anything except to evaluate your measurements. Just 'cause Oehler could put an SD calculator on his chronograph, now everyone thinks it has some relationship to load quality.
All that ever matters is; one, is it going fast enough, and two, is it accurate enough. Enough for what you say? Enough for the application at hand. Bench rest competitors want it accurate enough to have group sizes measured in the hundredth's of inches. Hunters gotta be able to hit the boiler room of whatever critter they are shooting, at the range they choose to shoot. (Which is more dependent on the shooter than the firearm in any case.) It has to be fast enoough to get there, wherever there is, with enough velocity to do what needs doing - punch a hole in the paper, knock over a steel silhouette, or penetrate, expand and kill the critter.
If you read enough gun articles, where the scribe slavishly records and comments on the SD, you will see, more often than not, that the lowest SD does not relate to the smallest group. Indeed, where the measurements are kept to evaluate the relationship between SD and group size, it can be demonstrated that there is no relationship.
Others clearly will disagree.