Abstract: The delivery of web content over an IP based network is accelerated by utilizing an intermediate node to receive requests for ML files, analyze ML files and create multipart objects that compound two or more browser's objects referenced by browser's links in the original ML file. The ML file is modified by replacing browser's links with multipart URLs that represent the multipart objects. When a subsequent user requests the page, the modified ML page is sent to the user and the multipart objects are more quickly and efficiently loaded into the user's browser.