A shame Robin isn't readily available to doscuss it, but I PROMISE you that these speakers weren't in the slightest bit "cobbled together" and that much listening, at our KJ branch amongst other places, was carried out in conjunction with measurements.

I think that presence dip is as much to do with the age and development of the drivers used back then - these things really have improved in recent times I believe. What you need to do is a computer model of the crossover slopes in conjunction with the driver characteristics to see how the crossover works and whether anything can be done without replacing either driver (or both). The fact that the off-azis performance in the review above is so good means that Robin got the phasing right (he's a great mathematician) - he'd had years fighting the LS3/5A batches he'd made for Audiomaster and was quite used to selecting components etc to get each batch through testing (KEF had tremendous variability issues with those drivers back then and every other batch would fail on forst testing - a bloody nighmare, the 3/5A).

You may just find that a modern tweeter will have far greater bandwidth at the lower end of its range and Wilmslow Audio may possibly be able to help. This really is where the DIY aspect comes in, together with computer simulations of the final results - someone here should be able to point you in the right direction..

Hope you can make something great out of something that's still good