Why in merge replication, tables at the publisher have identity range constraints ?

Hi All,

In merge replication what is the value of putting identity check constraints on the publisher tables ?

What is the risk if the publisher identity range is assigned at the end not the beginning, even it shouldn't care about the identity range for the server or the publisher ?

I know the identity ranges assignment is like this :

                   Range      Range
Server             Minimum    Maximum
------------------ ---------- ---------- 
Publisher                 1      100
Subscriber1             101      200
Subscriber2             201      300

but what if it like below ? and after each sync allocate new identity range for the subscribers, start with the maximum id from the publisher and allocate identity ranges for the subscribers that are bigger than the publisher maximum id

                   Range      Range
Server             Minimum    Maximum
------------------ ---------- ---------- 
Publisher               infinite      infinite
Subscriber1             101      200
Subscriber2             201      300
My question is: Any idea why replication team implemented it this way not the other way (i.e. with no constraints on the publisher) ?