Frame failure, seatstays at seat cluster

See other failures under 000.html.

Frame developed "scratchy" noise during a ride. Inspection following ride showed left seatstay cracked through, right seatstay cracked through except "cap" separating but not cracked through.

Notes:

Brazed steel Fuji "Club Fuji". Marked "Fuji VALite quad butted steel tubing 1769". Probably 1984. Frame weight unknown, but bike not especially light and original retail estimated under US$400.
First owner rode estimated 3,000km; weight approx 75 kgf. Rode hilly/enthusiastically, roads only.
Second rider weight approx 65 kgf. Rode hilly/enthusiastically roads and trails.
Almost all riding in dry weather. Almost all riding without a rack.
Belt-drive (non-electronic, non-reset) odometer attached by first owner. Registered under 18,000km at failure in 2005.
It appears the crack started at the "root" of the joint between seatstay and seat collar. The cracks spread across the stay until reaching the "cap". The left cap braze separated; ultimately, the left cap fractured. The right cap braze separated but the cap had not fractured when the failure was discovered.
Stay loads here are principally compression, so this failure seems odd. One hypothesis is the frame was brazed, then clamping the seatpost applied bending to the stay. Another rider remarked this is not a rare failure mode for stays attached this way.

The following frame failure, looks similar. From Yellow Jersey [http://www.yellowjersey.org] as of 2018/03/31.

Initially, it appeared that just the left stay was failed, but on closer inspection both were cracked.

As in the Fuji frame above, the frame tube is cut diagonally and capped with a flat plate. Cast stay tips may be more laterally flexible and so less prone to fail.