The Ferm classification was designed to be empirical: that is, based on actual samples of rock. We wanted to make sure the core-logging manual for each region included all the rocks that would be encountered by drillers, and illustrated them according to their frequency of occurrence. To create a representative collection of specimens, deep cores from the region were sampled from top to bottom, and a sample was taken from every distinct lithologic layer, no matter how many times that rock type occurred. This resulted in thousands of 6-inch-long rock samples.

Next, the rocks were sorted into broad groups based on easily observed properties corresponding to base lithology names such as sandstone, shale, and limestone. These groups were sorted into arrays to assess the variability of specific observable properties. For example, if the property of color in fine-grained rocks was of interest, all homogeneous mudstones were laid in a row with the darkest specimens on one end and the lightest on the other. If a sample had the same color as another already in the array, it was placed above that specimen. The resulting two-dimensional arrays looked like a histogram, in which the most common colors formed peaks in the array. Low-frequency positions along the arrays often suggested natural boundaries for rock classes. This method was repeated for all properties eventually used to define the categories of the classification. Properties assessed were grain size, mineral composition, color, fabric, bedding, structure, and fossil content.

A two-dimensional array for assessing frequency of occurrence of a single property.
A two-dimensional array for assessing frequency of occurrence of a single property.

Placing samples in a class was relatively easy when the full array was visible. But loggers would need to be able to make these decisions without the aid of comparative arrays, and we needed to be sure the classes were well chosen to minimize errors in judgment. Therefore, we conducted trials in which several loggers were given samples to classify according to proposed categories. If differences among the operators exceeded 20 percent, the boundaries were adjusted and the trial was repeated. We also found that providing guide samples near the class boundaries greatly increased repeatability.

Once all the properties were assessed for variability and groupings were agreed upon, the property classes were consolidated into lithologies that form the framework of the Ferm rock classification.


Last Modified on 2023-01-05
Back to Top