^^^ Back when I used to run elemental analysis on an ICP we'd put a QC check every so often and a blind, spiked sample randomly placed in the sample tray. Knowing those samples were in there kept me on my toes. Any good lab will do that. I am very surprised that TAMU gives unreliable results. Just for giggles, next time you send them a sample tell them that you have some dupes but don't tell them which.

On your end, part of what you need to do is thoroughly homogenize your sample. there are lots of methods of soil analysis. The method I used was (IIRC) 20 ml of hot nitric acid on 2 grams of dry, ground soil to extract P, K, Ca, Fe, Cu, Mg, Mn, Mo, B, S, and Zn ... I think that's all. put the slurry into a sealed teflon vessel and microwave it at 400 watts for something like 15 minutes. Then put it on a shaker table before filtering. We were after total extraction or all soluble elements -- everything except silica, pretty much. each sample was run through an ultrasonic nebulizer and squirted into the ICP.

Each sample was duplicated 3 times in extraction and then each duplicate was split into 2 vials for the ICP run.

You would think that with all of that duplication and QC's and random checks and rigorous attention to detail that we would have nearly identical results. Well, ... I had to re-run an a few samples because of spurious QC results or wildly different pseudoreplicates. It's the nature of the samples, methods, and analytical tools.

If I were you (anybody), I would take extreme care in sampling. Thi is how I did it: remove the leaf litter and expose the soil over about a ft^2 area. Use a soil probe or trowel and scoop a small amount of soil from each of 3 different holes within that 1 ft^2 area. Put the 3 samples together in a large baggy and mix the heck out of the sample, taking care not to let sweat drip into the bag (changes the nitrogen and sulfur content). Make sure all lumps are thoroughly broken and mixed. Remove pebbles and concretions (those lime clods that resist breaking). You now have enough soil for probably 1,000 samples. Take 3 samples from your mixing bag of an ounce or so and put them in 3 baggies. Label the baggies with some random labeling (which you record). repeat that process with a clean mixing bag for each sample.

If your area is homogeneous in soil type then you only need a minimum of 3 separate samples (of 3 pseudoreplicates each). If your area is more variable then you need a minimum of 3 samples (of 3 pseudoreps) per sub area.

By doing this, you will have enough samples to get a mean and a standard error for each area. That will tell you how much variability you have in each area. Also, the fertilizer recommendations can also be averaged and a standard error derived. This will tell you how much variability you can play with in your fertilizer rates.

Also make sure you ask for not just NPK and calcium, but also soil micronutrients.

It is expensive, no question. But you don't need to run the detailed sampling & analysis every year. Next year just take a single sample from each area and run that. You already know how much variability is in each area so you wont have to check that again.

Bot