I don't have my copy of "The Bicycle Wheel" here to check for quotes from that, but here is what he has to say about tied and soldered wheels in the rec.bicycles.tech FAQ:
----
While writing "the Bicycle Wheel", to conclusively determine what effect tying and soldering of spoke crossings in a wheel had, I asked Wheelsmith to loan me an untied pair of standard 36 spoke rear wheels, on on Campagnolo low and high flange hubs. I had an inner body of a freewheel machined with flats so that a wheel could be clamped into the vise of a Bridgeport milling machine while the left end of its axle was held in the quill.
With the hub rigidly secured, with its axle vertical, dial gauges were mounted at four equally spaced locations on the machine bed to measure rim deflections as a 35lb weight was sequentially hung on the wheel at these positions. The deflections were recorded for each location and averaged for each wheel before and after tying and soldering spokes.
The wheels were also measured for torsional rigidity in the same fixture, by a wire anchored in the valve hole and wrapped around the rim so that a 35 lb force could be applied tangential to the rim. Dial gauges located at two places 90 degrees apart in the quadrant away from the applied load were used to measure relative rotation between the wheel and hub.
Upon repeating the measurements after tying and soldering the spokes, no perceptible change, other than random measurement noise of a few thousandths of an inch, was detected. The spokes were tied and soldered by Wheelsmith who did this as a regular service. The data was collected by an engineer who did not know what I expected to find. I set up the experiment and delivered the wheels.
----
Todd Kuzma
Heron Bicycles
LaSalle, IL
http://www.heronbicycles.com/