Up ben at the edges so my issue with the seven countries study is not having measured diet in individual which was based on an understanding in nineteen sixty as i said nutritionist being mired in complexity in feeling that you really didn't learn anything if you didn't know all of these details which by the way we didn't really know how to put together so we're gonna put it together justice in as a nutrient on the other hand in the pure study in their diet papers they have not shown it by cohort so the the between this had to be very careful. Interpreting the between cohort ifferences are very interesting and within cohort. Differences are very interesting. And you shouldn't be ignoring either. A to shouldn't just collapse them. I wish in the seven countries study that there have been more individual data and there is actually in some of the cohorts some of that's been published and on the other hand. I wish in the pure study that Because they do have a substantial amount of individual data on every person they have all these different courts. I wish they would presented by court so that that you could look for a horse in situations statistic particularly low economic development countries. That are just completely different. Yeah one of the this. This is actually perhaps pulling in some of maybe your earliest work. But if i remember correctly did you work on some of the modeling on inter individual variation and blood cholesterol levels and higher that meant that it could make associations where blood cholesterol is mediation impacts Se dice it could make associations more difficult to detect out a population level. That actually goes back to a paper that was published in nineteen seventy nine thirty. Yeah getting to be ancient history That was my excursion into what is now called either regression dilution bias or errors in variables models in epidemiology. We assume that the independent variables are fixed. So if i'm take take the diet say. Diet patterns gore's had traded fat. I assume that we know what each person's value is and then the random variable is the long term outcome or the time to event and In the errors in variables models. You say well. I don't actually know exactly what that person. And.