Most times, SSL is enabled for the whole website, but depending on "which part" of the website your browser requests, your browser stays on the SSL version, or gets politely asked by the server to go to http instead, and then forms the same request on http again.
With "part" I mean the path part of the url you requested.
Two example traffics
First the not logged in request to the main site (cropped for better readability):
GET / HTTP/1.1 Host: amazon.com
Server sees the main site is being requested by someone not logged in, in https and sends the response:
HTTP/1.1 301 Moved Permanently Location: http://amazon.com/
Now the request for your login url:
GET /ap/signin/and_so_on HTTP/1.1 Host: amazon.com
Server sees the login site being requested, performs the login, and sends the response:
HTTP/1.1 200 OK (site content)
(successful login, no redirect)