25 miles: 55:53           26.83 mph
50 miles: 1:52:17         26.71 mph
100 miles: 3:55:09        25.52 mph
12 hours:  251.23 miles   20.94 mph

I did a regression of log distance versus log time on these data, and
got the following fit:

ln distance/miles = 3.3231 + 0.90097 * ln time/hours

(note: the semilog slope fitting the 25-100 miles would be less ...
the points are not exceptionally well fit by the model... nevertheless...)

Thus, if I predict his 24 hour and 6-minute distances, I get:

6 minutes: 3.49 miles
24 hours:  486 miles

The model predicts if you increase the duration of the exercise by a
factor 240, you get an increase in distance of 240^0.9.

So, for example, if you can ride 2.00 miles in 6 minutes (20 mph), then
you can ride 15.9 mph for one hour, and for 24 hours, you get:

(2 miles) * 240^(0.9) = 278 miles

If instead you can go 20 miles in 1 hour, for 24 hours you get:

(20 miles) * 24^(0.9) = 349 miles

Of course, your mileage may vary....

Dan