A nice theory is one side of the medal, however, in day to day practice there are much more hurdles to jump over, before accurate and reliable results are obtained.

The following, only seemingly exotic, example shows, what that means. It asks the question:

How accurate are precise data?

Observations

I have analyzed the raw data of three TRITON generated data files of approx. the same lenght (~ 11.5 h) of an Ames Nd standard sample. Data acquisition parameters are said to be identical.

Files:

• F4 and F12 obtained from J. Schwieters (Thermo, Bremen)
• P12 obtained from G. Caro (IPG, Paris)

The following figure shows a plot of the ^144/146Nd raw ratios for these three files:

Using the La Jolla standard value (R_N = R^144/146 = 1.3852), one observes a remarkable static offset of the measured raw ratio r^144/146 for all three files :

F4:	3.917 ‰	}
F12:	3.852 ‰	} per mass unit
P12:	2.477 ‰	}

This static offset e is defined as

r_N(0) is the raw ratio of the i₁₄₄ and i₁₄₆ ion currents at the very beginning of the data acquisition. The question is:

Has this offset any influence on the fractionation corrected ”true” ratio?

I will offer two answers:

The so-to-speak "automatic" (or implicit) correction of (mass dependent) static discriminations as a (favorable) side effect of fractionation correction, as shown in this section:

The (questionable) explicit correction of raw ratios before the application of fractionation correction formulae, as shown in this note:

Summary:

High precision runs seem to require a lot more checks and still not really proven correction methods, before their high precision may result in serious high accuracy results.

(July 2004)