Technical Abstract: Landscape responses to applied treatments have traditionally been accounted for in statistical analyses with blocking criteria (slope position, etc.), to compartmentalize variance and allow more accurate definition of treatment effects. However, within-unit distributions of soil properties can contain useful information about soil boundaries, patch structure, and applied treatments. This study compared distributions of soil bulk density (BD), total carbon (C) and nitrogen (N) in 4 experimental paddocks along a topographic gradient of an upland site in central Oklahoma in 2005, after 26 yr (1978-2004) of applied management. Included were grazed, conventionally tilled winter wheat, and 3 forms of management applied to tallgrass prairie; light stocking rates over long periods, high stocking rates over short periods, and no management (relict). In each paddock, soil cores to 25 cm depth in 3 segments (0-5 cm, 5-10 cm, and 10-25 cm) were collected at 1.5 m intervals along 150 m transects situated between a common ridge and toe slope. Split-moving window (4.5 m and 12.0 m scales) techniques were used to identify potential boundaries along transects. Soils in the wheat paddock were denser and had less C and N than soils in native paddocks. Differences among native paddocks were confined to the upper 10 cm, with denser soils under the high stocking rate and more C and N under the lighter stocking rate. Bulk density showed multiple boundaries/paddock that represented potential combinations of 3 to 5 common and/or locally distributed soils. Grazed paddocks had 2-3 definable C and N patches, while the relict unit had 7. These structures indicate the organization of soil properties had both common, landscape-scale features related to slope or parent material, and localized features potentially related to plant communities or management.