See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/297710669

Theory of sampling (TOS) - Fundamental definitions and concepts

Article · February 2015

CITATIONS READS

2 2,942

2 authors:

Kim H. Esbensen Cooper Wagner

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Working on the development of new sampling technology for powder blends and planning to attend world Conference in Sampling and Blending (www.wcsb8) in
Perth, Australia May 9 - 11, 2017 View project

TOS for NIR a.o. spectroscopic tehniques View project

All content following this page was uploaded by Kim H. Esbensen on 03 May 2016.

The user has requested enhancement of the downloaded file.

SAMPLING COLUMN

Theory of sampling (TOS)—

The Chinese sage Confucius is claimed This second Sampling Column intro- process, allowing a complex sampling
to have stated: “Speak precisely—and duces the most important fundamental task to be broken down into its indi-
wars can be avoided” definitions and principles of the Theory vidual stages and to apply individual, or
of Sampling without which no rational any required combination of, SUOs to
understanding and appreciation can be be able to cover all sampling situations.
established. We begin with a typology TOS’s sampling unit operations will be
of lot dimensionality, a term defining described in full detail in future sampling
both the geometrical dimensions as well columns. TOS focuses on the sampling
as the effective number of dimensions process and not the sample itself. Once
involved in sampling. extracted, there is no possibility to evalu-
Samples are extracted for various ate whether a specific sample is repre-
reasons, using many different sampling sentative of the target lot, or not. The
procedures in a wide range of application sampling process is the only guarantee
fields addressing a bewildering array of for a representative sample. Disobeying
different material types. One would think or compromising TOS’s principles will
that many potentially different sampling unavoidably lead to non-representative
Photo Wikipedia procedures would be needed. However, sampling procedures, which only will
the main purpose of sampling is the lead to “specimens” instead of “samples”
same—to be able to extract a small mass (distinction to be fully defined).
For the record, despite extensive search- of the target lot that is to be characterised This column presents a fundamental
ing with today’s many internet facilities, (analysed), i.e. to obtain a sample, which set of TOS’s terms and definitions without
we have been unable to confirm this accurately and precisely represents the which it is not possible to fully under-
attribution with certainty. Be that as it lot (see definition below). Sample repre- stand, nor discuss TOS in a meaningful
may, this is very powerful insight: this sentativity is therefore the sole criterion way.1 It has been found convenient to
dictum addresses what is needed for that must be honoured in order to be start out by defining “lot dimensionality”.
effective communication, and for the able to draw valid conclusions about the
relationship building thereupon and the characteristics of the original lot, while Lot dimensionality
meaning is clear: communication must non-representative samples (termed Lot dimensionality is characterised
at all costs avoid uncertainty, impreci- “specimens” in TOS) will result in a risk by specifying the number of effective
sion, vagueness in oral, written, tech- of erroneous decisions and conclusions dimensions that need to be covered
nical, scientific communication. Exactly without any possibility of knowing to by the sampling process. This approach
the same holds for those who want to what degree this is the case. allows definition of one-, two- and three-
communicate in science, technology The Theory of Sampling, the only fully dimensional (1-D, 2-D and 3-D) lots as
and industry, especially concerning a comprehensive approach to representa- well as the “zero-dimensional” (0-D) lot.2
topic that traditionally has been consid- tive sampling, allows a complete analysis Figure 1 compares TOS’s four cases of lot
ered “difficult”—and of this claim there is of representativity regarding all sampling dimensionality.
probably only a very few better examples methods, procedures and equip- The concept of lot dimensional-
than the Theory of Sampling (TOS). It is ment. Based on a set of scale-invari- ity becomes clear, for example, when
crucially important to be able to speak ant sampling unit operations (SUO), considering an elongated material
with the outmost precision. TOS defines sampling as a multi-stage stream, such as the case of dynami-

22 SPECTROSCOPYEUROPE www.spectroscopyeurope.com
 VOL. 27 NO. 1 (2015)

SAMPLING COLUMN
strata, stacks, stockpiles, silos) which
should, wherever possible, be trans-
formed to comply with a 1-D sampling
situation. In practice, this is often possi-
ble by locating a situation where the lot
already is in transport. Sometimes even
original 0-D lots are also transformed into
the desired 1-D configuration, because
this offers unbeatable optimal sampling
conditions (later column). Lot dimen-
sionality transformation constitutes one
of the governing principles (GPs) of the
TOS.
The reason for being this specific
about lot dimensionality is the inher-
ent (complex) heterogeneity of all natu-
rally-occurring materials, which makes
sampling far from a trivial materials
handling issue. Proper understanding of
the heterogeneity phenomenon, its influ-
ence on the sampling correctness and,
most importantly, how heterogeneity can
be counteracted in the sampling process
require a certain level of knowledge. The
purpose of the Sampling Columns is to
gradually build up this knowledge. The
first instalment of definitions follows.

Sampling terminology—
the tower of Babel
Lot
The complete entity of the original mate-
Figure 1. Lot dimensionalities: 0-D, 1-D, 2-D and 3-D lots. Potential increments are marked in rial being subject to sampling, e.g. truck
grey. Note how extracted increment can be made to cover the transverse dimensions only for load, railroad car, process stream, ship’s
1-D and 2-D lots but not for 3-D lots. The special 0-D lot is defined in full in the text.
cargo, batch etc. The lot (also termed the
sampling target or decision unit) refers
both to the physical, geometrical form
cally moving material on a conveyer and so are far from covering both the and size, as well as the material charac-
belts (Figure 2). This lot can loosely transverse lot dimensions fully (i.e. width teristics of the material being subject to
be described as one-dimensional, and thickness). This “covering” aspect is sampling.
since one dimension of the physi- a fundamental issue for TOS.
cal geometrical aspect dominates (the By contrast, a cross-stream cutter (a Heterogeneity
conveyer belt transportation direction). sampling device especially designed for Heterogeneity is the prime charac-
According to TOS, however, it is essential elongated material fluxes, which shall be terisation of all naturally occurring
to consider how the specific sampling extensively illustrated in one of the future materials, including industrial lots.
method applied is interacting with the sampling columns) will cover the entire Heterogeneity manifests itself at all
effective number of dimensions during depth and width of the stream, thereby scales related to sampling for nearly all
the sampling process. Employing grab fully reducing the sampling lot to one lot and material types. The only excep-
sampling (extracting a single increment dimension, i.e. the longitudinal dimen- tion is uniform materials, 3 which are
as the “sample”) on such an elongated sion of the material stream. See Figure 2 such a rare example that no gener-
material stream, which is a widely- for clarification of the practical aspect of alisation about them with respect to
applied but fundamentally-flawed extrac- lot dimensionality definition. sampling can be made here. TOS differ-
tion method, would in reality make this a According to the TOS, 1-D lots repre- entiates between two types of hetero-
3-D lot not a 1-D lot, since grab samples sent the optimal sampling situation, geneity, one concerning the spatial
are most likely only taken from the top preferred over 2-D and 3-D lots (e.g., distribution (distributional heterogene-
surface part of the moving material flux, industrial, geological or environmental ity, DH) of the target material and one

www.spectroscopyeurope.com SPECTROSCOPYEUROPE 23
 VOL. 27 NO. 1 (2015)

SAMPLING COLUMN

Figure 2. Lot dimensionality. Upper left: 1-D lot; right: 2-D lot; lower left: 3-D lot. TOS outlines that the principles behind representative sampling are
scale-invariant (see text).

referring to the compositional differ- a sampling target by itself. It is only the Elimination of the sampling bias is the
ences between the individual “units” sampling process, which can be termed first obligation for any sampling process
of the target material (compositional representative or not. in order to be correct.
heterogeneity, CH). The next sampling
column will deal extensively with these Correct sampling Reproducibility
two types of heterogeneity. TOS uses this term to denote that efforts Sampling variance, after removal of
have been executed, that has resulted in sampling bias, a.k.a. imprecision.
Sample successful elimination of the so-called
Correctly extracted material from the lot, “bias-generating errors”, a.k.a. the Increment
which can only originate from an unbi- Incorrect Sampling Errors (ISE). Incorrect Correctly delineated, materialised unit
ased, representative sampling process. sampling errors will be discussed more of the lot which, when combined with
The term sample should always only be in one a future sampling column. other increments, provides a multi-incre-
used in this qualified sense of “represen- ment sample. This procedure is termed
tative sample”. If there is doubt as to this Representativeness “composite sampling” in TOS, with the
characteristic, the term “specimen” (see Representativeness implies both correct- result being a “composite sample”.
below) should be used instead. ness as well as a sufficiently small
sampling reproducibility (sampling vari- Composite sample
Specimen ance). Aggregation of several increments, the
A “sample” that cannot be documented number of which is designated as Q.
to be the end result of a bona fide repre- Sampling bias A composite sample represents “physi-
sentative sampling process. It is not Systematic deviation between the aver- cal averaging”, as opposed to arithmetic
possible to ascertain the representativ- age analytical sampling result and the averaging of analytical results from indi-
ity status of any isolated small part of true lot concentration, a.k.a. accuracy. vidual increments.

24 SPECTROSCOPYEUROPE www.spectroscopyeurope.com
 VOL. 27 NO. 1 (2015)

SAMPLING COLUMN
Sub-sample sampling tool. The group size depends For a full set of necessary and suffi-
Correctly mass-reduced part of sample on the sampling tool (mass/volume) cient definitions, referral is made to the
(primary, secondary...). A sub-sample is and the sampling process as well as how horizontal sampling standard DS 3077
a result from a dissociative (disaggrega- the tool is implemented and operated. (2013).4
tion) process; a composite sample is a
result from an integrative process. Scale Notes and references
The principles described by TOS are 1. Currently there is a debate on “Alternatives to
Gy’s Sampling Theory?” on LinkedIn. This is a
Fragment scale-invariant, i.e. the same principles good example of the critical need for precise
Fragment refers to the smallest separable apply to all relevant scales and stages speaking. We shall return to this discussion
unit of the material that is not affected by in the sampling pathway (lot, sample, in these columns but, significantly, after all
the necessary basic concepts, definitions
the sampling process itself (e.g. particles, sub-sample). and principles have been properly intro-
grains etc.). By naming the smallest unit- duced.
of-interest a fragment, TOS is also able to Zero-dimensional lot (0-D lot) 2. The special case of a zero-dimensional lot
refers to a lot that can be effectively, mixed,
treat the situation in which the sampling The 0-D lot is characterised by display- moved and sampled throughout with
process results in fragmentation of some ing no internal correlations between all complete correctness. Usually these are
of the original units. potential increments, thus opening up for small lots, which can easily be manipulated.
A full definition of the 0-D lot is given in the
relatively easy practical sampling. A 0-D definition section of this sampling column.
Group lot can be manipulated—at least in prin- 3. Uniform materials: Materials with a repeated
A number of spatially correlated frag- ciple—for example, by mixing or direct in (correct) sampling reproducibility lower than
2%. Such materials do not occur naturally
ments, which act as a coherent unit toto splitting, the work necessary may (exception gasses and infinitely diluted solu-
(increment) during sampling opera- vary significantly as a function of the lot tions etc.).
tions. In practical sampling, the only mass, ML, but also by of other relevant 4. Representative Sampling—Horizontal
Standard. Danish Standards DS 3077
group of interest is the actual increment features, e.g. stickiness, irregular fragment (2013). www.ds.dk
being extracted, i.e. the material in the forms.

continued from page 17 9. P.C. Burns, “U 6+ minerals and inorganic 16. J.B.M. Novo, F.R. Batista, C.J. da Cunha,
compounds: insights into an expanded L.C. Dias, Jr, and F.B.T. Pessine, “Time-
structural hierarchy of crystal structures”, resolved studies in hydrogen uranyl phos-
2. S. Cot ton, Lanthanide and Actinide
Can. Mineral. 43, 1839–1894 (2005). phate intercalated with amines”, J. Lumin.
Chemistry (2nd Edition). Wiley, Chichester
doi: http://dx.doi.org/10.2113/gscan- 124, 133–139 (2007). doi: http://dx.doi.
(2006).
min.43.6.1839 org/10.1016/j.jlumin.2006.02.009
3. A.Y. Vasiliev and M. Sidorov (Eds), Uranium:
10. R.L. Frost, “An infrared and Raman 17. S.J. Formosinho, H.D. Burrows, M.Da
Characteristics, Occurrence, and Human
spectroscopic study of the uranyl micas”, Graca Miguel, M.E.D.G. Azenha, I.M.
Exposure. Nova Science Publishers,
Spectrochim. Acta A 60, 1469–1480 Saraiva, A.C.D.N. Ribeiro, I.V. Khudyakov,
Hauppauge, NY (2012).
(2004). doi: http://dx.doi.org/10.1016/j. R.G. Gasanov, M. Bolte and M. Sarakha,
4. E. Faulques, R.E. Russo and D.L. Perry,
saa.2003.08.013 “Deactivation processes of the lowest
“Raman spectral studies of uranyl sulfate
11. N. Sanchez-Pastor, A.J. Pinto, J.M. Astilleros, L. excited state of [UO 2(H 2O) 5] 2+ in aque-
and its urea complex structural isomers”,
Fernandez-Diaz and M.A. Gonçalves, “Raman ous solution”, Photochem. Photobiol. Sci.
Spectrochim. Acta A 49, 975–983 (1993).
spectroscopic characterization of a synthetic, 2, 569–575 (2003). doi: http://dx.doi.
doi: ht tp://dx.doi.org/10.1016/0584-
non-stoichiometric Cu-Ba uranyl phos- org/10.1039/B300346A
8539(93)80216-W
phate”, Spectrochim. Acta A 113, 196–202 18. E. Hashem, J.A. Platts, F. Hartl, G. Lorusso
5. E. Faulques, R.E. Russo and D.L. Perry,
(2013). doi: http://dx.doi.org/10.1016/j. I.M. Evangelisti, C. Schulzke and R.J. Baker,
“Raman studies of uranyl nitrate and its
saa.2013.03.094 “Thiocyanate complexes of uranium in
hydroxy bridged dimer”, Spectrochim. Acta
12. R.L. Frost, J. Cejka and G. Ayoko, “Raman multiple oxidation states: a combined struc-
A 50, 757–763 (1994). doi: http://dx.doi.
spectroscopy study of the uranyl phosphate tural, magnetic, spectroscopic, spectroelec-
org/10.1016/0584-8539(94)80013-8
minerals phosphuranylite and yingjiangite”, trochemical, and theoretical study”, Inorg.
6. T.R. Ravindran and A.K. Arora, “Study of
J. Raman Spectrosc. 39, 495–502 (2008). Chem. 53, 8624–8637 (2014). doi: http://
Raman spectra of uranium using a surface-
doi: http://dx.doi.org/10.1002/jrs.1868 dx.doi.org/10.1021/ic501236j
enhanced Raman scattering technique”, J.
13. R.J.P. Driscoll, D. Wolverson, J.M. Mitchels, 19. M. Mohapatra, B. Rajeswari, R.M. Kadam,
Raman Spectrosc. 42, 885–887 (2011). doi:
J.M. Skelton, S.C. Parker, M. Molinari, I. M. Kumar, T.K. Seshagiri, N.K. Porwal, S.V.
http://dx.doi.org/10.1002/jrs.2926
Khan, D. Geeson and G.C. Allen, “A Raman Godbole and V. Natarajan, “Investigation
7. R.L. Frost, J. Cejka and M.L. Weier, “Raman
spectroscopic study of uranyl minerals from of uranium luminescence in SrB4O7 matrix
spectroscopic study of the uranyl oxyhydrox-
Cornwall, UK”, RSC Adv. 4, 59137–59149 by time resolved photoluminescence, ther-
ide hydrates: becqerelite, billietite, curite,
(2014). doi: http://dx.doi.org/10.1039/ mally stimulated luminescence and elec-
schoepite, and vandendriess-cheite”, J.
C4RA09361E tron spin resonance spectroscopy”, J. Alloy.
Raman Spectrosc. 38, 460–466 (2007).
14. http://rruff.info/Metauranocircite/R070721 Compd. 611, 74–81 (2014). doi: http://
doi: http://dx.doi.org/10.1002/jrs.1669
and http://rruff.info/ Metauranocircite dx.doi.org/10.1016/j.jallcom.2014.05.096
8. G. Guimbretiere, A. Canizares, P. Simon,
Ibidem /Saleeite; Ibidem/Ulrichite 20. S. Tsushima, C. Gotz and K.F. Dr,
Y.A. Tobon-Correa, M.R. Ammar, C. Corbel
15. G. Meinrath, S. Lisc, Z. Strylad and C. “Photoluminescence of uranium(VI):
and M.F. Barthe, “In-situ Raman observa-
Noubactep, “Lifetime and fluorescence quan- quenching mechanisms and role of
tion of uranium dioxide weathering exposed
tum yield of uranium(VI) species in hydro- uranium(V),” Chem. Eur. J. 16, 8029–8033
to water radiolysis”, Spectrosc. Lett. 44,
lyzed solutions”, J. Alloy. Compd. 300–301, (2010). doi: http://dx.doi.org/10.1002/
570–573 (2011). doi: http://dx.doi.org/10.
107–112 (2000) . doi: ht tp://dx.doi. chem.201000408
1080/00387010.2011.610857
org/10.1016/S0925-8388(99)00739-2

www.spectroscopyeurope.com SPECTROSCOPYEUROPE 25

View publication stats