[Walt Fricke]

CRNT918

I've never seen such a recommendation.

Perhaps you are thinking of the Brey-Krause advice that a harness bar not be used for shoulder belt attachment. That is, unless it is the nifty truss design which they also sell and is designed for those loads. The BK pitch is that standard harness bars are not adequate to do more than serve as guide bars for the shoulder harness loads - which is why they also sell a piece which they call a guide bar.

A properly installed bolt-in cage or roll bar (proper ones, not cosmetic ones like the Jeep folks sometimes favor) is plenty strong enough to serve as a harness attachment. I think welding in is a better mousetrap, but we are talking questions of degree, not yes or no.

For years, nobody paid much attention to the length of shoulder harnesses from shoulder to attachment. Lately, most have awoken to the fact that the longer the belt, the more stretch in it during a severe impact. While some stretch is good because it greatly reduces peak G loads on the body, too much is bad because it allows body parts to impact parts of the car. And the peak loads from that secondary impact are nasty indeed.

So now it is common to see recommendations that the shoulder harnesses be mounted as closely to the shoulders as can be. Which means to a horizontal cross bar in the main roll hoop, is positioned - as it should be - at or just below shoulder height.

Running the belts over a guide bar and back to the package shelf, or down to the floor pan, or back to the rear seat belt attachment points, is not optimum, as all leave quite a length of belt to stretch. I don't think anyone would fail tech with this, but it is somethng folks should be aware of.