Good advice and I will take it into account.
After looking at many of the Namban Rengo runners' splits from the Tokyo Marathon, I decided to figure out my Napa splits per 5k rather than just miles. (By the way, the constant pattern was slower from 25k, even slower from 30k and much slower from 35k. Well, this is hardly news, but it shows that all of us suffer from a combination of not enough long/hard training and setting our sights (pace) too high. It just shows that one or both of those factors HAS to be addressed in order to run a good (and enjoyable) marathon. Maybe setting reasonable goals is the approach more likely to succeed...like I should have targeted 3:05 instead of 3:00.)
Anyway, my 5k splits were:
21:16 / 21:23 / 21:41 / 22:12 / 21:56 / 22:20 / 23:09 / 24:21 / 11:28 — 3:09:58
Now here's what's interesting. Here are my splits from Tokyo one year ago:
21:18 / 21:19 / 21:58 / 22:15 / 22:26 / 22:32 / 23:03 / 24:00 / 10:58 — 3:09:53
And Tokyo two years ago:
23:12 / 21:31 / 22:06 / 22:56 / 22:48 / 23:01 / 23:43 / 23:54 / 10:20 — 3:13:35
So I did not improve this year, and compared to two years ago, my last 7k is much worse!
Which I guess means there is something to be said for starting out slowly.
One year ago, my goal was 3:06 (22:05/5k). I started too fast, but was on pace through 35k.
This year, my goal was 3:00 (21:15/5k) and I was on pace for 5k — obviously an unreasonable goal. However, I was again on pace for a 3:06 through 35k. Maybe if I had started more slowly I could have finished better (or maybe not - my legs were really in bad shape).
But how about pre-race training volume?
In the 7.5 months before Napa Valley (from start of training to start of final 2-week taper) I ran 2,644km (average 352km/month).
In the 7.5 months before the other two, I ran 2,034 and 2,035 (271km/month).
So the extra 600km did nothing for me. Also, in the 7.5 months before Napa, while I of course did more long runs, I did fewer "hard" runs (don't have time to work out the exact numbers, but looking at the log that seems to be the case).
(Another factor is that a lot of the distance was in July, August and Sept., after which I fell off a lot, which is far from the ideal pattern.)
So what does that mean? I think it means that first, an extra 80k a month (from 67 to 88k a week / 41 to 54 miles) is not enough to make a difference. I don't think even 100k a week would make a difference. I would have to get up to, as Joachim and others have said, at least 120k/73miles a week. However, I cannot run that much without injury. The three months I did averaging 115k/week resulted in my knees starting to shred (sharp pains followed by constant soreness). Even though ALL of that was at an easy pace.
The second conclusion is that 67k a week (as in the previous 2 years) even with hard runs, is also not enough.
So that leaves me with finding a mean at 90k/55 miles a week, and doing as much of that as possible at a hard pace. Which is what I came up with in the last post!
And I agree with Steve that I will have to work into that, not just start banging out hard 90k weeks.
I also had a brainstorm and modified the previous training schedule. The problem is that having to do intervals every Wednesday night makes a 7-day cycle very difficult. My previous ideas were 2 days off a week or 3 days off a week. 3 off would make the other 4 days too tough, but I would like to get an occasional 2 days in a row of rest. So the answer is: a 14-day cycle! To wit:
Wed - am 12k, pm 8k Easy Hard
T - 15 H
F - Off
S - 15 H
S - 25 H
M - Off
T - 15 E
W - 12, 8 E H
T - Off
F - 15 H
S - 15 E
S - 30 H
M - Off
T - Off
This results in 7 hard runs, 4 easy runs and 5 rest days. Also no more 30+k runs, which I used to do a lot of. The hard components have to be worked out, but "hard" does not mean blasting 15k from beginning to end. Workouts like 5 moderate, 5 x 1000m, 5 moderate.
The first week is 90k, the second week is 80k. And every 4th week I will cut back the last two runs to 10 and 15k, so that week will be 60k. That gives me 320k per 28 days, or about 340k per month, which is good but not excessive, I think.
This won't start until mid-May, but I love to plan ahead.