Reliability: An experiment is reliable if, when it is repeated, it gives the same result. Note that many teachers (and students and tutors) seem to think that reliability can be improved by repeating an experiment. This is only true in one case - when the lack of reliability is due to poor experimental technique. However, assuming you perform the experiment correctly, redoing it will establish whether it is reliable, but it will NOT improve its reliability. (It wll, however, improve accuracy.)
For example, titrations are a reliable experiment - if done correctly, they give basically the same answer each time.
Accuracy: An experiment is accurate if the result it produces is close to the correct result.
For example, a correctly designed titration is accurate, as it will give the correct concentration of a solution.
Continuing with titrations as an example:
- if you use an inappropriate indicator, the experiment will remain reliable, but will cease to be accurate.
- if you wash the conical flask with the solution it is to contain, the experiment will no longer be accurate or reliable - this is because you will get the wrong answer, and different runs will give different answers, depending on how much of the washing solution is left behind.
Now, consider the heat of combustion of alcohols experiment from Production of Materials. This experiment is reasonably reliable - if everyone in your class did it, they would produce similar answers, albeit with a bit of spread due to random errors. However, the experiment is not accurate, as there is a large systematic error (heat loss to the environment) leading to answers being consistently far too low.
You could modify this experiment to adjust for heat loss, by calibrating the equipment. This would improve the accuracy. You could also use a draft shield to help reduce variation due to wind, etc. This would improve the reliability.
When answering questions on this stuff, you MUST discuss both aspects, and remember that some modifications will improve accuracy only, some will improve reliability only, and some will improve both.
