I'd do it exactly as "only" suggests. 3 mi warmup, 4 mi tempo, 1 mi warmdown. Warm into the tempo portion from below.
There's a deep structural problem, it seems to me, with the whole idea of calculators using race results to prescribe certain paces not just that one might run in races of longer duration, but that one should use as a guide to training paces--"half marathon pace," "marathon pace," and the like.
Here's the problem. The calculator works from whatever race result you put into it. Usually we work from a race result that we obtained on a good day--when we were well-trained, tapered, ready to go.
But I never run my tempo runs on those legs. I always run them untapered, in the middle of a training week. Sometimes I make the second half of a two-hour run into a de facto tempo run at half-marathon EFFORT, based on HR and a general memory of what half-marathon effort is supposed to be.
When I do that, though, I end up running a pace that's quite a bit slower than the pace I actually run my half-marathons at. For example, on two successive weekends I ran the last 63 minutes of a 2:05 run at 89% HR max (174 of 196), and averaged 7:52 pace. I was working at half-marathon pace effort. But the half marathon I actually ran four weeks later I ran at exactly 7:00 pace. Because I was rested, it was a cold perfect day, I was adrenalized.
Now, if I'd tried to run 7:00 pace for the last 7 miles of my 15 miler, after 8 easy miles of warmup--a strenous but not unprecedented "half marathon pace tempo run"--on the training day in question: well, I'd have been running more at 10K effort. I'd literally have run myself into the ground.
It's worth remembering this when you're working a tempo run and not quite keeping up with what you're "supposed" to be running. "Supposed" is an illusion. Sometimes, of course, it's important to do some running at the actual pace you hope to maintain, since economy at that specific pace is a trained-in thing.
But remember what Pfitz suggests: start your long runs at 20% slower than MP and finish at 10% slower. If your imputed (calculator-ed) MP is 7:30, he suggests that you start at 9:00 and finish with 8:15s. Why? He explains that 8:15s, in training, pretty much develop the specific muscles (and thus economy) that 7:30 pace will require on race day--where as LSD mileage, at 9:00 pace, doesn't do that.
7:30 -------> 8:15 is a 45-second spread. That's a lot. Sort of like my 7:52 -------->7:00 spread.
What Pfitz DOESN'T say, but could have, is that 8:15 pace at the end of a long run probably puts your HR pretty close to actual marathon pace HR, i.e., the HR you'll be able to hit at 7:30 on race day, when you're fully tapered, glycogen-stuffed, adrenalized, etc.
Sorry to hijack the thread. This has just been on my mind for a while. I haven't seen it discussed, and it should be.
http://This message has been edited by KudzuRunner (edited Jun-16-2006).