I see this posted multiple times, & since it references peer-reviewed papers (inconveniently without links), it merits investigation. I’ll read every paper listed, where publicly available, look for further work responding to it or related to it, & comment on its relevance to the original claim.

There’s a lot of material to go through, so I make no commitment about when I’ll finish. This page will be updated as I make progress.

Here goes:

Autistic Disturbances of Affective Contact


A first description of autism cases by a medical community just discovering this condition.

This has many citations (Google Scholar counts 8389 at the time of writing). This is understandably so – it describes a troubling condition. The fact that this is the oldest study cited also tends to contribute to its citation count.

With regard to the topic, there is little mention of vaccination, & one of the subjects of the study (Paul) suffered from vaccine-preventable illnesses.

There is a mention, in the case of Richard, of having symptoms after a smallpox vaccination at 12 months. It should be noted the inoculation was after the onset of autistic behaviours, not before.

Now, let’s start on the claims enumerated by the article.

Hepatitis B Vaccination of Male Neonates and Autism


Boys who received hepatitis B vaccinations of a type discontinued in 2002 in the first month of life had an increased likelihood of autism by up to 3x.

Causality unknown.

In total, 86996 boys’ histories were considered in a cross-sectional study spanning birth years 1997 to 2002 that compared those vaccinated against Hepatitis-B in the first month of life against those never vaccinated against it, & attempted to adjust for confounding factors.

A significant result was found – that boys born before 1999 with a vaccination record, who received the first dose of hepatitis B vaccine during the first month of life were 3.002 times more likely to receive an autism diagnosis than those vaccinated later or not at all. This is what is mentioned in the paper’s abstract.

The 95% confidence interval for this result is 1.109 to 8.126. That is, the authors are 95% confident that the true value is between 1.109 and 8.126, and believe the most likely value is 3.002.

There’s a fair discussion of the findings provided by the authors, and suggestions for more research into vaccination scheduling. Since the background to the the original claim is that vaccines are harmful, it is relevant to print the last sentence here:

Our findings do not suggest that the risks of autism outweigh the benefits of vaccination; however, future research into hepatitis B vaccination scheduling is warranted.

Porphyrinuria in childhood autistic disorder: Implications for environmental toxicity


While an interesting study on the effects of heavy metal poisoning, this paper neither mentions, nor is relevant to, vaccination.

It could be imagined that the relevance to vaccination is that Thimerosal or Thiomersal (a preservative used until recently in many vaccines, now less popular due to unfounded public mistrust) is a chemical that has a mercury atom.

Mercury is a toxic heavy metal, so would certainly qualify as an environmental toxin. The dangers of heavy metals are usually due to the body’s poor ability to rid itself of them, & so a slow build-up to a dangerous dose occurs.

Fortunately, chemistry can produce profound changes in the behaviours of atoms when they are combined in molecules. For example, while chlorine is a dangerous greenish gas that was used to devastating effect to kill soldiers in the first World War, attaching a sodium atom to it produces sodium chloride (AKA table salt).

In the same way, despite the fact that mercury is a dangerous poison, capable of slowly building up in the body (it reduces by half in about 120 days), Thiomersal, while also toxic, decomposes into ethylmercury readily in the blood. The body can get rid of half all its ethylmercury in roughly 18 days – almost 10 times faster than inorganic mercury. This is still dangerously cumulative, but much less so that inorganic (neat) mercury.

There has been no link yet found between Thiomersal and autism spectrum disorder (ASD).

The main source of mercury for people is dietary. Vaccines are not the source of heavy metal poisoning you’re looking for.

Theoretical aspects of autism: Causes—A review


A large quantity of literature was reviewed. Multiple causes for autism have been identified, & are suspected to overlap. The association of vaccines to autism was limited to implication & hypothesis.

I.e. there might be a link between vaccinations & autism, but we don’t know if there is one.

There is an autism epidemic. A great deal of work is being done to understand the epidemic, & this document provides a summary up to 2011. Included, is also a substantial description of the effects of ASDs & other similar effects, such as heavy metal poisoning.

Much content of this review was devoted to the work on vaccinations. It makes mention of the coincidence of changes to vaccine regimes & increases in autism prevalence. Other than temporal association, the only other potential link between vaccinations & autism is the fact that Thiomersal contains mercury, & the effects of mercury poisoning include the chief symptoms of autism.

Uncoupling of ATP-mediated Calcium Signaling and Dysregulated IL-6 Secretion in Dendritic Cells by Nanomolar Thimerosal

Dendritic cells are sensitive to Thiomersal. One cited article references autism in its title. No further references to autism. While it’s possible dentritic cells are involved in autism, nobody knows how autism works.

The only link I can imagine the author of the original claim might make from autism to this paper is that autism is suspected to be linked to immune system abnormalities, & dendritic cells are a part of the immune system.

It could be the case that Thiomersal contributes to autism. Still no evidence, just more calls for more research, which is being done.

Abiogenesis – the bar

Abiogenesis gets a lot of stick from sceptics because it seems intuitively wrong. In my experience, the most common problem people seem to have with the idea is that it appears to rely on a highly complicated chemistry arising from nothing more than random processes. While there isn’t yet a conclusive empirical description of how this may have occurred, there is work in the area, and it doesn’t seem to be in any way in violation of the physics and chemistry we know today.

Disclosure: I totally buy abiogenesis. I think it’s not even slightly implausible. I have seen order come from chaos in a number of cases, not least in an evolution demonstration I wrote for fun last year. This in no way violates the 2nd Law of Thermodynamics, in the same way that tidying cutlery in a drawer doesn’t.

There’s also a point to be made that supernatural hypotheses, such as the Christian creation myth, are also examples of abiogenesis, but that’s nitpicking, and I won’t talk about that here.

One thing that I do credit sceptics with, is that they have a point about probability. I think they don’t have the point they believe they have though. Typically the trope goes like this, “the probability of a fully-formed protein (or even in some cases, a working cell!) emerging from a bunch of atoms is astronomically low, therefore, it didn’t happen”. My thoughts on this idea are:

  1. Because of the continuing work into the subject, and the lack of information from the development of life, I think we don’t know how to tell the probability of a self-replicating molecule forming that could kick-start the process of evolution.
  2. The probability of a replicator developing doesn’t need to be high to start life, because it only has to happen once.

This second point is the crux of this post. What would be an acceptable probability for abiogenesis to be a 95% probable explanation for life on Earth? I think this could be very low.


I wonder what the probability of abiogenesis in a given litre of reactive substrate would have to be to produce a 95% likelihood of development of life in our galaxy.

Drake equation

Let’s borrow some values from the Drake equation to get some ballpark numbers to estimate locations where life could develop.

Symbol Meaning Current Value
f_{p} Fraction of stars that have planets 1
n_{e} Average number of satellites suitable for life 0.4

Let’s add to that some more numbers:

n Number of stars in the galaxy 100000000000
V Average volume of liquid water available for chemistry (cubic km) 1386000000

I took n from the Drake equation current values, and V from here, assuming that Earth has a typical value.

So that gives us an available volume of:

n \times f_{p} \times n_{e} \times V \times 10^{12} litres

Now we need to know how long this was able to react. Let’s take an educated guess from Universe Today, and pretend the early universe was too violent for life formation until the galactic disk had formed. That gives us 10 billion years to play with:

T Duration of life development (billion years) 10
A Age of developed life (billion years) 3
P Probability abiogenesis will occur in the galaxy in the available time 0.95
P_{a} Probability abiogenesis will occur in 1 litre of substrate in 1 year ?

This gives:

P_{a}=\frac{P}{((T - A) \times 10^{9} \times n \times f_{p} \times n_{e} \times V \times 10^{12})}


P_{a}=\frac{0.95}{((10 - 3) \times 10^{9} \times 10^{11} \times 1 \times 0.4 \times 1.386 \times 10^{9} \times 10^{11}}

Which is very very small. Very small.

2.4 \times 10^{-42}

On average, then. If abiogenesis is rare enough that it only happened once in our galaxy with a confidence of 95%, and a lab scientist was performing an experiment with the right experimental conditions on 1 litre of substrate, she would have to wait about 3000 billion billion billion times the age of the universe so far. On average! If she were unlucky, perhaps she’d have to wait longer…

Evolutionary Cline Simulator

I recently had time to code a biological cline simulator.

It runs locally on your machine, and is written in JavaScript. For this reason, don’t expect usable performance from phones and tablets!


Here’s the result of one run after a few thousand steps.

evolution run showing cline

Selection Pressure
The selection pressure image describes the selection pressure within the environment. In a red region, a red coloured organism will do well. In a green region, a green organism will do well. Of the 3 colours in the RGB colour system, blue is non-selective. This means a purple (blue + red) organism will do just as well as a red one, and a turquoise one (blue + green) will do just as well as a green one. Regions in black cannot be entered by the organisms.

All genes are expressed. No alleles are recessive.

The environment canvas is where the organisms are rendered. They are rendered by their phenotype. For this model, their phenotype is simply an RGB colour. The difference between the organism’s colour and the colour of the corresponding pixel on the selection pressure image gives a value of how well the organism is faring – the best possible value would be a perfect match, i.e. a difference of 0.

For simplicity, organisms are hermaphroditic. They move randomly a short distance each step. Whenever they encounter each other, either by landing in the same pixel, or in one of the 8 adjacent pixels, they mate.

The simulated organisms are diploid, and are capable of being heterozygous, receiving one chromosomefrom each parent. Each chromosome is implemented as an RGB colour. The chromosome has 3 genes (red, green, blue), each with 256 alleles (potential values of 0-255).

Organisms are generated in a random location within a perimeter, and all have identical genomes.

The organism’s phenotype is calculated by averaging both “chromosomes”, producing a colour halfway between each chromosome colour.

If the environment is unfavourable for an individual (for example, a green organism in a location in the environment that corresponds to a red pixel in the selection pressure image), it typically moves further each processing step than if it’s in a favourable environment. This gives organisms a very limited quantity of behavioural “intelligence”. They are more likely to spend more time in a favourable environment, than an unfavourable one.

Matings produce 0-1 offspring – 0 offspring if the two parents’ gametes are incompatible, otherwise 1.

Each parent organism generates a single gamete. A gamete is a single chromosome. Gametes are generated by randomly selecting 3 genes from the parent’s available 6, 1 each of red, green and blue. There is a chance that a small mutation may be applied to each gene at this point, altering 0-3 alleles by a small amount.

Genetic compatibility is measured by comparing the difference between all 3 genes of each of the parent’s gametes. If the gametes are sufficiently similar, 1 organism is produced by the mating. If not, 0 organisms are produced – the 2 potential parents are too genetically different; they have speciated.

Injury and death
Each organism cannot live for more than 1000 processing steps. Invariably, the environment, and/or competition with their peers kills them before old age can.

Organisms have a value of “health” when this reaches 0, the organism dies. Every processing step, the organism’s phenotype is compared with the selection pressure of its pixel, and the difference (modified by a “hostility” factor), subtracted from its health.

To prevent population growing exponentially, there is a small health hit each time an organism encounters another organism; in highly populated areas, this has the effect of shortening organism’s lifespans, and preventing a population explosion. This could be imagined as multiple animals competing for resources, and spreading diseases with each other.

An organism’s health can only ever stay stable and decline. There is no way an organism can increase it’s health value.