Hi all,
I would like to establish what shop bought test kits give a reasonably accurate result?
I'l start with dissolved gases, O2, CO2, NH3, in these cases there aren't any kits that will work, and even meters costing ~£1000 are only going to give you a ball-park figure. We can use indirect techniques (like a drop checker), for some of these parameters, although we may need to be able to measure another parameter (in this case carbonate hardness (dKH)) fairly accurately for this to work.
After that it really depends upon the reagents in the kit, and whether they are appropriate for the conditions in the water tested.
It isn't really a case of "this test works" or "that one doesn't", nearly all of them will give a reading in appropriate conditions, but what you can't do is say that 10ppm NO3 really is 10ppm, or that a similar reading in different conditions (acidic rather than alkaline water, chloride ions present etc) really are the same.
If you think of it a bit like a colour printer, it might be great with primary colours, but once you get it on to muted browns, greens and greys, they all print the same colour. If I was going to buy kits, I'd buy semi-quantitative kits which include standards and probably a methodology which includes 2 or more reagents.
The major problem for dissolved compounds is that whatever reagent you use, and whether you use photometry, spectrophotometry or colourimetry, you nearly always need to run a blank (everything else the same, but none of the element (or compound) you are interested in present) and standards (known values) to create a standard curve, and then run your samples (with some dilution, replication and re-testing), before plotting the standard curve and samples and deriving your unknown values from your known. Even then you may not have a high degree of repeat-ability, and this means that you have a low level of confidence in your method.
Scientists quantify this by using "95% confidence intervals", this just means that you have a zone where you have a 95% likelihood of the true mean (or median) lying. I say mean or median, because confidence intervals are symmetrical around a mean, but this assumes that the data follows the normal distribution. If you use the median value of a data set we don't need to know the nature of the data distribution, but our confidence interval may not be symmetrical. Most modern analytical kit does the curve drawing etc for you, but you still have an answer with some degree of uncertainty (this is why we use probablility values for significance). Have a look at this for NO3. <
http://www.epa.gov/etv/pubs/01_vr_nec_fntk.pdf>, if you look through the "verified EPA testing methods" there are reviews of available kits (in the USA) for a wide range of compounds.
Titrimetric methods, where we add an acid to a base, (or vice versa), and use a colour pH indicator to indicate the end point, suffer from the same problems, although they work quite well for dKH, where we are dealing with water with a large amount of dissolved carbonates.
PH we know can be accurately tested especially with a probe.
I think pH is a very tricky measurement, mainly because you need a measure of dKH before you can interpret it, it is a log10 scale, and it is ratio so tells you nothing about amounts. Really you are back to the colour printer again red (acid) or blue (alkaline) fine, shades of green are more problematic.
Conductivity meters really do work as accurate repeatable dip meters, but they don't record a very useful parameter.
cheers Darrel