In attempting to decide whether the two sets of results derive from the same population, we must essentially decide whether the two data sets appear to follow the same distribution or not. We can do this by looking at the parameters associated with that distribution and comparing the figures we get for the two data sets.
So, first we must decide on what type of distribution our shared population might follow. We don't know anything about our population, so we must try and provide a 'model' to which we could say it approximately follows. A good one to start with (as it's probably the most common) is a Normally distributed population (I haven't done any of the calculations, so I'm not sure whether this the correct one or not). I'll not go through all the aspects associated with this distribution, but essentially we say that if we plotted the entire population of values on an axis, it would follow a bell-shaped curve pattern, centered at some mean value and having some standard deviation from the mean. As the shape is symmetric , the population mean would be equal to the population median.Here's what it would look like:
http://www.globalspec.com/reference/69565/203279/11-5-the-normal-gaussian-distribution
If we were to plot our values in the two data sets individually (eg; boxplot or histogram), we would probably find that they might not look exactly like this smooth curve. This is because we have relatively small sample from a population in each case, so it's difficult to get a precise judgement compared to a population model which would theoretically contain infinitely many values. So, perhaps the best 'numerical' way of identifying whether they derive from the same population is if the 'Chemical' and 'Atmospheric' data sets have similar sample mean and sample standard deviation values (meaning that they are random draws from a Normal population with a similar mean and standard deviation).
In doing this, we essentially use these sample parameters to estimate the parameters of the population that each of the data sets derive from. If the estimates are similar, then there would be good reason to believe that the samples are from the one Normally-distributed population.
So, the first thing to do is to calculate the mean and standard deviation values associated with each of the data sets. There are many other things such as calculating and comparing 95% confidence intervals for the population means of both Chemical and Atmospheric or performing hypothesis tests...I'm not going to into these methods as there a bit specialised, but if you know about how to do them when encountering similar problems before, then perhaps you could use them here too. What I'd suggest to do is as follows:
-Compare the mean and median values of each of the two samples individually; if they are in good agreement with one another then the populations which the samples derive from are most likely to be Normal.
-Compare the mean and standard deviation of each data set; if they appear to be in good agreement with one another, then the populations are most likely the same. If not, then they derive from two distinct Normal distributions with population parameters which are not the same.
Hope that helps!