Weissmann was the first experimenter who recognised the importance of the error rate. And he decided to measure it with Q-beta. And a funny story was, we had our winter seminar... we will later certainly talk still about our winter seminar, but I might take this ahead of time.  He came to one of our winter seminars and said, 'Manfred, I have an experimental result which must interest you'. In fact he had two experimental results. The first one, he said is, 'I measured the error rate'.

[Q] What system did he choose?

Q-beta... phage Q-beta. With a total phage and with an infection. Here was not Spiegelman's experiment, but he infected coli cultures, and he said, 'Well, it has a very high error rate, about a million times higher than, let's say, a coli cell'.

[Q] Quite a sophisticated system. Coli is quite sophisticated.

Yes, it's a living organism and there's viruses dependent on a host cell of course. But, that's funny, it's a million times larger. So in viruses we should expect much higher error rates than in other organisms. And the second result he brought out was that he said, 'This system doesn't consist of a given sequence. It is a distribution of sequences'. How could he find out about it? He had a population of virus particles and then he took out a small volume element where only one virus particle is in, and cloned it, and grew up a population of each single virus particle. And then he measured the sequence of it and he found, in the different clones, they are all different sequences.  No sequence agreed with the other, there were a few errors but no one... So I said, 'There is no wild type. The wild type is an average of all these sequences'. And I came to this seminar and said, 'Charlie, listen to my lecture. I have a theory which first tells you that viruses must have an error rate which is a million times larger than coli', which meant... it's not quite a million times, about a million, but it's more than a million times compared to human cells for instance. So the error rate which Weissmann had found was, I think, 3 x 10^-4. That means about every three thousandth building block is wrongly copied. And since the Q-beta has about four thousand odd, it means in each replication round you make at least one mistake. And these mistakes are necessary for evolution, for adapting it. That's the first result. My theory told me the error threshold is exactly of that order of magnitude, namely that the product of the length of the virus times the error rate is of the order of magnitude of one. So you have to make mistakes in order to adapt to new... but you must not make too many mistakes then you lose your information. Was exactly... we calculated from my equation, and got exactly his number was... what he found experimentally.

The second result, that he has a mutant distribution in which no sequence agrees with any other...

[Q] 100%.  Does not agree by 100%.

Not by 100%, of course, the average sequence must be the same otherwise you lose the information. But, I say, that's exactly our quasispecies model.  The quasispecies tells yes, if you are at the error threshold, your reproduction is so that you never make a completely exact copy.