Just released an updated version of our 'aqp' package for quantitative soils investigations, available on CRAN and R-Forge. Some of the major changes are listed below:

 from the aqp NEWS file:
-------------------------- aqp 0.98-1 (2010-11-23) --------------------------
* added basic demo: demo(aqp)
* soil.slot() now computes aggregate probabilities over user-defined segments
* re-write of weighted profile aggregation functions
  - aqp package now requires Hmisc package
  - using Hmisc::wtd.{mean,var,quantile} to compute values
	+ better adjustment of weights when computing weighted SD
	+ fixes long-standing bugs with wt. mean/SD when NA present in x_i and not in wt_i
	+ weighted quantiles now computed
* addition of new (experimental) S4 classes and methods
  - these are now in the aqp_S4 branch
* added n.depth.ticks option to profile_plot()
* profile_compare() algorithm stabilized, new default settings
* removed code in profile_compare() that is now obsolete
  - note that this will cause changes to numerical classification of soil profiles

Just for fun, I tried applying our numerical soil classification algorithm to a database of 455 soil profiles-- part of the SoilVeg soils collection. Calculation of the pair-wise dissimilarity matrix required about 2 minutes on a 5 year old computer, and about 150 Mb of RAM. The resulting dendrogram representation of between-profile dissimilarity was far too large for a standard sheet of paper, so scaled versions are included below. Six classes were generated by cutting the dendrogram, and soil properties used to generate between-profile dissimilarity estimates were aggregated by group. The final figure below illustrates median depth functions for soil properties within each group.

AQP Demo: SoilVeg Dendrogram

AQP Demo: SoilVeg Dendrogram Zoom

AQP Demo: SoilVeg Group-Wise Depth Functions