From the article:
Normal – 44.5%
Italic – 44.9%
Bold – 45%
Bold Italic – 43.8%
[...]
Whilst Bold did the best I’m not convinced that it’s
something that it worth investigating further with that
page.
I feel like the author has a weak statistical background, which is unfortunate when running a site that is, essentially, all about applying statistical tests. The difference between Italic and Bold is basically indistinguishable without a sample size of a few hundred thousand.
Even telling Bold Italic and Bold apart reliably (i.e. lying outside each others approximate 95% confidence intervals) requires a sample size of more than 6000.
(Admittedly, my statistics isn't awesome, and it's late here.)
(Also, www.quicklytest.it and quicklytest.it are acting differently. Specifically, the former is saying "The requested URL <url> was not found on this server." for every request.)