It is quite the amazing trend line, and I have done similar ones myself. There is no observable slow down in observable value generation since the Gen Re acquisition in 1998.

But I think it's important to remember that the long term trend line really ought not to be extrapolated. It has happened, and it's remarkable, but there is no particular reason to think it should or will continue at the same pace. There are some quite good reasons to expect it to slow down, for that matter, as well as reasons to think that at some point (though not yet) real book per share will cease to be a "good enough" yardstick of fair value.

A small geeky note as well: the trend line as drawn could not have been known in advance. This is the fatal flaw (among many) of the BMW charting method. Rather, each point should be the current point on the trend line for all the data to date, so there is no crystal ball effect. This gives a very different, much curvier line, with overshoots and undershoots. It's a fairly small problem with Berkshire's real book per share, but something to be aware of. This emphasizes the point that the trend line known up to any given date (such as today) should not really be extrapolated. When you see the variation in the "trend line to date" levels in the past, it makes the concern clearer.

To deal with both those issues, personally I prefer to simply smooth the recent data. If the growth rate slows, the smoothing line slows, with a modest lag.
e.g., the blue line in this graph.
http://www.stonewellfunds.com/SmoothedRealValue.pn...
Each point on the blue line is calculated solely from data available prior to that date, and does not use any data more than four years old (using a 16 quarter WMA). The amount of smoothing can be varied to taste--I choose just enough stiffness to be consistent with my theory that true value per share does not actually drop during bear markets (drops in book per share), while staying as close as possible to the recent observed data points.
(note, on that particular graph the yellow spots are not real book, but the higher of [real book] and [96% of peak-to-date real book]---dips have been capped)

But back to the first comment: it really is a remarkable history of steady growth in value. The trend line fit is amazingly good!

Jim