We have a bulleted list that reads:

5 Effects:
(5.1) if T is bool or char_­type, initializes value with v; otherwise,
(5.2) if T is char and char_­type is wchar_­t, initializes value with static_­cast<wchar_­t>(v); otherwise,
(5.3) if T is a signed integer type ([basic.fundamental]) and sizeof(T) <= sizeof(int), initializes value with static_­cast(v); otherwise,
(5.4) if T is an unsigned integer type and sizeof(T) <= sizeof(unsigned int), initializes value with static_­cast(v); otherwise,
(5.5) if T is a signed integer type and sizeof(T) <= sizeof(long long int), initializes value with static_­cast(v); otherwise,
(5.6) if T is an unsigned integer type and sizeof(T) <= sizeof(unsigned long long int), initializes value with static_­cast(v); otherwise,
(5.7) initializes value with handle(v).

The otherwises are all the end of the bullets. Normally, we put the otherwises at the beginning of the next bullet. I don't think there's a reason for this case to be different - and it kind of reads like we're unconditionally initializing value with handle(v).