Reason being is because that for the diagram if all of the little dots are connected another single dot then it is a pure but if one little dot isn't connected to another one then it isn't a pure substance.
The density of CO2 getting from experiment is 0.1/0.056 = 1.79 g/L. The percent error of this is (1.96 -1.79)/1.96*100%=8.67%. So the approximate percent error is 8.67%.