I suspect that the original enquiry might have referred to equipment of the 1950s and perhaps earlier 1960s. If so, then it might be worth looking at viewpoints of the time, before really good speaker analysis tools were available.
Gilbert Briggs, in his book “Loudspeakers” devoted a couple of pages to cone shape, as follows:
Briggs LS Cone Shape.pdf
The comments therein tend to support Edward’s position.
Also, pertinent to Leon’s comment about the Bush VHF64, Briggs also provided some comments on the benefits of parallel operation of close-mounted speakers:
Briggs LS Parallel Operation.pdf
It could be that Bush used a pair of dissimilar but same-sized speakers to increase the cone area without using a much larger diameter unit, to obtain the benefit of smoother bass performance from the mutual coupling, and, through different cone shapes, to obtain the desired mid-range balance.
When all is said and done, I imagine that final performance is more dependent upon well executed is the chosen pathway rather than the actual choice of pathway. (Although many hi-fi makers would try to persuade us otherwise.)
Cheers,