I think the difference is as a result of downloading a smaller file in case 1. In case 2, the combined effect of downloading multiple files is equivalent to downloading a big file.
In order to correctly measure throughput, you need to download a big file (eg 20Mb). Otherwise, the throughput will be fictitiously low. This is because the throughput starts increasing from zero(just before u start downloading), and if the file is small, it will finish downloading before you achieve the maximum throughput.
Hope this helps
Sheldon