To my mind, the entire length of the c.p,c, need to meet minimum requirements - there's no exceptions for short lengths whereever they are. That said, c.p.c.s can be shared between circuits and only have to meet the requirements of the most demanding one, so if you had several cables glanded into a plate an in effect several lengths of G/Y in parallel to the earth bar, you might only need to verify that the total c.p.c. was OK for the largest circuit rather than each one having to meet the entire requirement for its own circuit on its own.

The gland plate situation has always been a bit odd - if steel conduit is used instead of SWA, it's common not to have any extra earth tails at all, even where the conduit i used as the c.pc. Go figure.

   - Andy.

It very much depends what is acting as the CPC - in the case of a metal box and a decently "graunched" joint, it may be that nothing extra in the way of CPC is needed. Now It may be the sort of folk that do SWA vs the sort that do conduit, or perhaps more charitably,  the fact that conduit can't come in at a slight angle and twist after installation, but there seems to be much more  SWA working loose than there is loose conduit,

Personally I'm a great fan of the "Piranha" style nuts (or one of the variations on the theme) as opposed to banjo washers if there is anything requiring a positive CPC to armour- and plastic boxes / painted boxes etc are often in this category,

In such a case I'd expect the 'tail' of CPC to match the SWA core diameter, but if pushed I could be convinced that the 'half' rule could apply, but some justification would be needed for using a lot less.

That said if the armoured cable was a given size for reasons of length and voltage drop, but the currents were limited by MCB or whatever to a lower value, then a section of  thinner CPC or indeed a section of thinner phases and neutral cables for a short hop, may be justified, or even needed to fit the size of terminals or to route easily.

Mike

Actually there are 3 sets of 2x150 4c each with a separate 50mm CPC, following that logic if the banjos were doubled up and one piece of 50mm earth was used to bond each set to the earth bar internally within the switch board all three together would equal 150mm, therefore satisfying the half size rule. 
But at the load end within the distribution board one piece of 50mm, or even 2, one from each banjo would fall short and would not meet the half size rule. 
I did wonder about the cumulative effect of the CPC’s but from an engineering/ design standpoint people only tend to look at one element at a time. 
I also looked up BS EN 61439 and found some useful BEAMA guidance, essentially I think it demonstrates the metal enclosure withstand, presumably up to the same rating as the busbars which is 50kA. Going by the manufacturer data this would imply that once glanded there would effectively be no real requirement to bond the gland itself, however I know this approach has been frowned upon for a long time now, presumably there are regulations in BS7671 that stipulate the glands should be bonded. 
The issue is as I see it  is designers never bother to add the manufacturer details or CofC to the design with the cable calcs preferring to base assumptions on equipment sheets which are vague and lack detail, all round I think this issue is caused by a lack of understanding on many fronts. 

Actually there are 3 sets of 2x150 4c each with a separate 50mm CPC, following that logic if the banjos were doubled up and one piece of 50mm earth was used to bond each set to the earth bar internally within the switch board all three together would equal 150mm, therefore satisfying the half size rule. 
But at the load end within the distribution board one piece of 50mm, or even 2, one from each banjo would fall short and would not meet the half size rule. 
I did wonder about the cumulative effect of the CPC’s but from an engineering/ design standpoint people only tend to look at one element at a time. 
I also looked up BS EN 61439 and found some useful BEAMA guidance, essentially I think it demonstrates the metal enclosure withstand, presumably up to the same rating as the busbars which is 50kA. Going by the manufacturer data this would imply that once glanded there would effectively be no real requirement to bond the gland itself, however I know this approach has been frowned upon for a long time now, presumably there are regulations in BS7671 that stipulate the glands should be bonded. 
The issue is as I see it  is designers never bother to add the manufacturer details or CofC to the design with the cable calcs preferring to base assumptions on equipment sheets which are vague and lack detail, all round I think this issue is caused by a lack of understanding on many fronts.