Correspondence: Michael S. Christensen, Springhouse Skin Research, Inc., 215 Springhouse Lane, Merion Station, Pennsylvania 19066, USA. E-mail: mikesenior@verizon.net was then pressed against the medial cheek, causing the glue to spread to a thin film the width of the slide (1 inch) and approximately half its length (1.5 inch). The slide was left in place for 5 minutes while the cyanoacrylate hardened as it polymerized. Before removal, an outline of the slide was drawn with a Sharpie permanent marker, leaving a demarcated area on which a second cyanoacrylate-coated slide was applied. Two sequential surface biopsy samples were thus obtained, taken from precisely the same site, a procedure that caused minimal discomfort and erythema. Slides were placed face up on the stage of a stereomicroscope for quantitative assessment of horny casts and determination of mite density per square centimeter. A second biopsy is shown in Supplementary Figure 1. To quantify mite densities, with the slide in place the microscope stage was moved under the objective lens to generate five distinct representative fields. The casts in each were counted separately, and a mean of the five values was used to calculate casts per square centimeter. To liberate the mites from the casts, a drop of olive oil was placed on the slide, sufficient to cover the entire sample. Casts (a total of 20, randomly selected) were then disrupted individually with a fine needle. As each cast was disrupted, the number of mites released was immediately counted while the mites swam freely in the olive oil medium (Supplementary Figure 2). Mites remained mobile in the olive oil for nearly 3 hours, in contrast to less than 2 hours in mineral oil, before becoming motionless and more difficult to distinguish from cellular debris. Even when large numbers of casts were found on a biopsy, not all casts contained mites, so we determined two factors after counting: the percentage of casts infested and the mean number of mites per infested cast. Using these two factors and the number of casts per square centimeter, we calculated the number of mites per square centimeter The data and text commentary herein were drawn posthumously from labor atory notebooks and writings found in Dr. Kligman’s study and bedroom. Comments in brackets were added by Barbara A. Gilchrest (Department of Dermatology, Boston University School of Medicine) and James J. Leyden (Department of Dermatology, University of Pennsylvania School of Medicine). Demodex folliculorum was first described independently by two investigators nearly 170 years ago. Human beings are the sole host for this mite (Supplementary Figure 1 online), which resides exclusively in sebaceous follicles and has been repeatedly but inconclusively implicated in rosacea by more than 1,000 peer-reviewed papers, approximately half of them published since 1990. Sometimes visible within follicles in routine cross-sectional biopsy sections, Demodex is more reliably detected in “skin surface biopsies,” a technique first employed by Marks and Dawber (1971) and modified by Mills and Kligman (1983). The purpose of this investigation is to further refine the technique to make it more accurate, reproducible, and quantitative— characteristics necessary for implicating the mites in skin disease. [One of Kligman’s characteristics was to periodically start over and assume everything previously said or published was probably wrong, even if it was he who had said or published it. At times in the past he had dismissed the possibility that Demodex has a role in skin disease, but at the end of the trail he reversed his position and determined to seek data in support of the possibility. Thus, with his longtime collaborator, Mike Christensen, Kligman embarked on a bottom-up re-exploration of whether Demodex plays a role in skin disease. True to his lifelong approach, Kligman decided to begin by improving, if possible, the technique for quantifying Demodex.This methodology is described online.]