One thing that is especially troublesome in the original email is the need to analyze 200 mg of soil in order to generate 75 micrograms of N to get linearity. I own your standard Delta Plus XL, vintage 1999. We routinely, and I stress routinely, analyze standards with 10 micrograms of N and get perfectly good isotope ratios. If there is a need to push, then push we can, and often do. So, before trying to cram, shove, and pummel 200 mg of soil into the EA throat, I would have a look at why the mass spec is not linear unless it gets 75 mikes.

Second, Hillary's point about sample heterogeneity is very important. What information one might derive from looking at a low organic/N soil is a big question. There is no doubt that one can pick up different bits of organic and inorganic reservoirs at very low sample concentrations. What would the isotope ratios mean? and why are they important for you to measure? If there were not some really good serious questions, I would shrug and suggest that if N is that low in a rock/soil sample, it is unclear about its origin and interpretation.

The other issue being discussed is how one would measure the effects of a soil matrix on a small amount of organic matter. There is not an easy answer to that. We've had in our lab a couple of projects that demanded ultra low (non-ionprobe) measurements of carbon isotopes in rocks. Blanks are important--especially the blanks on tin or silver boats, the general level of carbon or nitrogen in lab air, etc. Tied up in experiments to determine isotopes on soils/rocks is considering what happens when a sample reacts in the combustion oven. I am no Bruno Lavettre, but I do understand that the reaction is flash combustion. My EA is not set up to flash combust 200 mg of anything, as we have a small oxygen loop installed. The question of basalt vs. quartz vs. other minerals is an important point as well.  The matrix for combustion is important.

Not withstanding all of these comments, I do commonly, but not routinely, run rocky bits for N isotopes and concentrations. Old Silverback John Hayes wrote a paper on signal to noise ratio considerations for continuous flow isotope measurements. Maybe signal to noise is what is causing non-linearity problems and the need for 75 micrograms. In my hands, it is signal to noise that typically "gets" us on the nitrogen isotopes at low levels. 

In summary, I'd go back to my original scientific question and see if it could be answered by another method, or I would modify my question if my technique did not satisfy it.

Marilyn