402 > plot(mymod)
Notice that the residual plot has at least two recognizable
components, but it couldn't help but look like this. There is perhaps
some suggestion that a function of temperature might be needed, like
perhaps 
, but don't be fooled. The places where there are big
residuals are for temperatures where there were both no failures, and
failures. I tried a number of additional functions of temperature,
and nothing much worked.