He is correct of course, but it’s not that simple, the trick is to match the directivity of the horn to the low frequency drivers at the crossover point. As with most things it’s a compromise.
FWIW RCF uses the same horn and crossover point in this box.
http://www.rcf.it/products/touring-and-theatre/tt5-a
If you have a look at the directivity plots of the TT5a what you will see is that the vertical control starts to collapse at around 2kHz.
If you crossed this horn over at somewhere between 1K5 and 2kHz all would be good regarding the horn directivity (in isolation) BUT the two 10” drivers would not combined very well, there will be problems associated cancelation between each of the two 10’s, and 10’s and the horn vertically. This occurs because of the different path lengths to the listener as you move up and down vertically.
If you can keep the 10” driver(s) and the horn all within ¼ of a wavelength at the crossover there would be no issue with the off axis response but in practical terms that is too low for the horn to work; if you can get close to ½ a wave length then it’s not too bad … ½ a wave length at 650Hz is 261mm, more or less the spacing of the 10’s and horn - in practice it seems to work 
The cost of crossing so low is that the horn does not load the compression driver very well at the lower frequencies, so you need a large robust compression driver capable of substantial output to keep up, hence the BMS. The RCF 950 or one of the new B&C drivers with 4" diaphragm would also work.
