Robert L. FOLK (1981) on Settling Tubes

Updated 14-Aug-10 04:01h

Petrology of Sedimentary Rocks by
Robert L. FOLK, 1981

Recently, I read the latest edition of the popular book by Robert L. Folk in the Internet. I was disappointed by about 8 lines of his text on Settling Tubes. I immediately called Bob and asked him to exchange his text with that which I had sent him by eMail. However, he replied that the geology librarian he had consulted was strongly against any change: “even if it is an electronic version, the date is 1980 without further emendations and it would be very non-kosher to change any part of it retroactively“. Of course, this was not my intention, I wanted him to write his current comment quoting me. But Bob encouraged me to publish my text independently of his book. I would like to make clear that a sand granularity has value for sedimentologic interpretations only if made extremely accurately and properly described. This requirement alone gives no chance to rapid sieving and poor (“inexpensive”) settling tubes.

Please compare his text (red print below) with my suggestion (green print farther below) and tell me.

Size Analysis by Settling Tube
by Robert L. Folk (1981)

For rapid, generally less accurate work, grain size of sand may be measured by settling the grains through a water column. Fancy devices are available to make sure all the grains start settling together. Best results are obtained with water columns at least 10cm. in diameter, and with a small number of grains. Results are recorded by continuous weighing of grains accumulating at the bottom, or by automatic recording of pressure differences. Both “sievers” and “settlers” are strongly opinionated as to which is the best method of measuring grain size -- one may just as well ask, “Which is the best way of doing carpentry - a saw or a hammer?”

Size Analysis by Settling Tube
by Jiri Brezina (2009)

For rapid and most accurate work, granularity of sand measured by settling of grains through a water column, can reach the highest accuracy from all sizing methods, if the following terms are fulfilled:

The above sample size is statistically representative and reduces an initial density streaming to acceptable minimum. For that purpose, the sample introduction device homogenizes the upper 5 cm of the suspension by eccentric rotation and vibration of the Venetian blind lamellae.

The continuous weighing at the bottom must be fast (>96 % of load recorded within <50 milliseconds) and sensitive (0.01 percent resolution). Such a balance must be well insulated from environmental vibration (the tube suspended on shock absorber. Suspension pressure differences are not suitable as they require high particle concentration.

The electronic weight output signal must be “high fidelity”, i.e. with a high S/N (signal to noise) ratio (>80 dB). In order to collect optimum number of measurements (weighing signal samples), the sampling scheme must follow at constant intervals of a logarithmic sedimentation velocity, 0.02 PSI. In order to reduce the signal noise further, the voltage must be digitized by a fast integrating 16 bit ADC (Analog/Digital Converter), delivering at least 1000 measurements per second. These data are averaged over the non-linear time intervals (a simple mathematical signal filter).

Meeting these requirements can never be expected from “inexpensive” settling tubes, such as those sensing a suspension pressure difference. European colleagues confirmed the words R. L. FOLK wrote 1962: "The fault, dear Friends, lies not in the sands, but in your tubes, that they are skewing things." The “inexpensive” settling tubes have discredited grain size interpretation, which has also suffered from taking values and/or “polarity” of unsuitable distribution parameters skewness & kurtosis as a magic.

Joe R. CURRAY (1960) in his pioneer work “Tracing sediment masses by grain size modes” demonstrated the best common sense interpretation of natural size distributions: he decomposed them into normal (Gaussian) components and used them as natural tracers. In his time without comput­ers, he worked graphically. Tjeerd Jerry van ANDEL (1973) could decompose by a special analog DuPont Curve Resolver. Unfortunately, no other sedimentologist had this expensive device available, DuPont stopped its production & support from economic reasons. Because the graphical decomposi­tion is time-intensive and inaccurate, the Curay’s idea could not be further used before J. Brezina constructed his advanced settling tube and adapted a stable decomposition program by Isobel CLARK (197?) for his MacroGranometer™ system.

Therefore, the most effective interpreta­tion of natural distributions must be based on the three parameters (mean, spread and component percentage or proportion) of each normal distribution component. In studies based on decomposition of the observed distributions into a few Gaussian components, the highest analytical accuracy is inevitable. The mixing of normal distributions should be understood as either of different materials or of different proces­ses.

In clastic deposits, do we really need grain size? A grain non-sphericity may so much reduce the hydrodynamically equivalent sphere size that the grain may become classified into a category sedimenting under another hydrodynamic regime. Instead of the grain size, we actually need the grain settling rate PSI, which eliminates the problem of particle shape.

A conversion of a normal (Gaussian) PSI sedimentation velocity distribution into a PHI grain size distribution generates a different standard deviation and negative skewness of the PHI distribution. Jiri BREZINA (1963) showed graphically that these changes result from the Kapteyn’s transformation. His original data, based on limited resources known to him (Russian publications), had to be replaced by more reliable ones as in his equation (1979). Both the changes (of SD and SK) depend not only on the PSI mean and SD values but also on the Shape Factor values used for the conversion.