Armchair Science vs Real Science
In the world of pseudoscience and quackery - and especially in the world of "alternative" autism therapies - there are a group of people who sit in front of television cameras and pontificate on what "might" cause autism and what "might" treat it. I think the term "Armchair Scientist" fits these people well.
As with "Armchair Quarterback", "Armchair Scientist" is not a flattering term.
An Armchair Scientist gets to sit and speculate about what might be without ever having to do the hard work of actually doing any research to find out what is. The average Armchair Scientist doesn't even bother to review the published literature - they just make it up from what people have told them, what they've read on the Internet and what comes to them in their dreams.
However, the more pernicious of the Armchair Scientists have a pasing familiarity with the literature and cite it to their benefit. Several good examples of Armchair Scientists can be found on the FAIR Autism Media website. My personal favorites from this site are the dynamic (definitely sarcasm - see the video) duo of Geier and Geier (pere et fils).
In their video clip from the FAIR site, Geier and Geier expound at length (in a performance that can only be described as "petrified" - it's too stiff to be "wooden") about how they "discovered" that testosterone is the linchpin to explaining:
 How thimerosal invaccines causes autism (their bread and butter - literally).
 How the myriad of disparate and illogical autism treatments "work".
 Why more boys than girls are autistic.
It's a masterful piece of work, to be sure, and it would be a magnificent addition to the corpus of medical knowledge except for one small problem...
They don't have any data.
They begin with a stirring description of how they "found" that mercury inhibited a key enzyme in the pathway (more properly, the web of interlocking synthesis pathways) from cholesterol to testosterone and how this discovery led them to the belief (unsupported, as yet, by any data) that this was the key to the autism puzzle. Their "discovery" was from a paper by Ryan RA and Carrol J, Studies on a 3beta-hydroxysteroid sulphotransferase from rat liver (1976).
In this paper, Ryan and Carrol describe that the enzyme 3beta-hydroxysteroid sulphotransferase (EC 126.96.36.199) was, "...inhibited by p-chloromercuribenzoate and HgCl2". However, if you do a search of Medline for that enzyme, you will find that it is regulated, stimulated and inhibited by a wide variety of substances besides mercury.
So, let's look for a moment at some of the other claims they make on this video.
They cite a "study" by Boyd Haley, PhD who reportedly exposed neurons in cell culture and found that testonsterone enhanced the toxicity of mercury. This would be interesting - if not earth-shattering - news, except that it hasn't been published anywhere. Boyd Haley has a long list of publications and his name alone was able to get unadulterated twaddle published in an otherwise respectable journal, so I doubt the lack of publication is due to the infamous "conspiracy of silence".
A possible explanation for this absence of publication is that Boyd Haley doesn't feel confident enough of his results to embarrass himself in front of his peers. It's one thing to show video of neurons "unraveling" to a group of credulous parents - it's another thing entirely to show the same piece of work to people who know the "tricks".
That said, there are several studies that have looked at the role of testosterone in mercury poisoning (which should never be confused with autism - the symptoms and signs are completely different). A study by Barregard, Lindstedt, Schutz and Sallsten, Endocrine function in mercury exposed chloralkali workers (1994), showed that "Serum total testosterone, but not free testosterone, was positively correlated with cumulative Hg exposure."
Another study by Tanaka, Naganuma, Miura and Imura, Role of testosterone in gamma-glutamyltranspeptidase-dependent renal methylmercury uptake in mice (1992), showed that "Renal mercury content in 4-week-old male mice was twofold higher than that of females and increased with age, but remained constant in females." and "Seven days after castration of 4-week-old male mice, both renal mercury content and gamma-GTP activity were decreased to the levels in females."
On the other hand, a study by Mohamed, Burbacher (a familiar name) and Mottet, Effects of methyl mercury on testicular functions in Macaca fascicularis monkeys (1987), showed "The MeHg-induced increase in semen abnormalities was not accompanied by any significant changes in serum levels of testosterone."
Even more interesting is a study by Hirayama,Yasutake and Inoue, Effect of sex hormones on the fate of methylmercury and on glutathione metabolism in mice (1987), which found:
"Twenty-four hours after oral administration of MeHg, urinary Hg levels were significantly higher in males than in females. Tissue Hg levels of males were higher in the kidney, but lower in the brain, liver and plasma than those of females. The fate of injected MeHg in castrated males was similar to that in normal females except for its brain levels. This feminization of the mercurial behavior in the castrated males was restored by treating with testosterone propionate (TP). When control mice were treated with TP, urinary excretion of Hg increased in both sexes, whereas renal Hg level increased only in females. Administration of estradiol benzoate (EB) to males decreased the renal accumulation and urinary excretion of Hg, whereas its hepatic levelsIn short, Hirayama et al [thanks Clone3G, for the reference!] found that testosterone protected the brain and increased mercury excretion. And that estrogen made the mice more susceptible to mercury poisoning. Quite the opposite of what Geier and Geier speculate might be happening in autistic children!
So, the connection between mercury and elevated testosterone - which seems so solid to the Geiers - is not clear at all, even at overtly toxic levels. Nor does the idea that testosterone impairs mercury excretion or increases mercury toxicity seem to be holding water.
The real laugh in the show was when the Geiers described how mercury "coordinated" the binding of testosterone into large "sheets" and "matrices" that were resistant to breakdown by enzymes. They based this - so they said - on crystallographic data. Since there is only one report of co-crystallization of testosterone and mercury [Cooper A, Gopalakrishna EM, Norton DA, The crystal structure and absolute configuration of the 2:1 complex between testosterone and mercuric chloride (1968)], it seems that is the source of this datum.
Apart from the fact that the processes used to crystallize testosterone (or most any other biological molecule) for X-ray crystallography are not happening in the body, there are some flaws in this "reasoning":
 If the mercury is complexed into a matrix of testosterone and mercury, it is not available to cause problems. It would seem that these "sheets" of testosterone would be an ideal way to sequester mercury away from delicate brain tissue, which would leave open the question of why the Geiers want to reduce testosterone.
 Mercury doesn't complex with carbon, oxygen or hydrogen very well - it "prefers" sulfur, but will complex with phosphorus or nitrogen in a pinch. below are the structures of several compounds related to testosterone (including testosterone itself). For those unfamiliar with the notation, the lines have a carbon atom at each end unless a letter is present.
Clearly, there is no high-affinity spot for mercury to attach.
 How is mercury - a brainless atom that cannot read structure diagrams - to know that it should attach to testosterone and not one of the several "look-alike" molecules? How many of the intelligent life-forms reading this blog can spot the testosterone molecule?
Here's the "key" for the above test:
So, we have no data to support the idea that testosterone binds preferentially to mercury, that mercury bound to testosterone (if it were possible) would be bioavailable (capable of doing harm). We also have no data that mercury causes an increase in DHEA (as proposed by Geier and Geier) or that testosterone impairs excretion of mercury. What's left?