See the Forensicator's blog, he explains the whole process in detail (relative modification timestamp differences) as well as explaining about the FAT filesystem anomalies discovered and other factors observed that this article and most articles critical of his research (or Lawrence's referencing of the research, to be precise) tend to omit in order to make a strawman out of transfer rate attainability.
Re: Re: Re: Also worth noting - compression and resumption
http://theforensicator.wordpress.com/
Re: Re: try again
There was no conversion mistake - as is being suggested.
Re: The answer is obvious
(We're talking a total batch of files at around 19Gb, of which the NGP-VAN archive's contents appear to be a subset of)