58th Annual STS (1998–1999)
Finalists
Leon Marcel Bellan

A Study on the Human Ability to Detect Soot Deposition onto Works of Art

Abstract

Due to the deposition of airborne black carbon particles in museums and art galleries, valuable artwork often is soiled. The soot particle deposition rates are already known for many museums, but knowing the levels of carbon particle coverage at which humans can detect image darkening is also important. Samples of soiled surfaces are characterized quantitatively and by human observers. From this data, the level of darkening at which the average observer can detect soiling is determined.

Other Info

The question that we intended to answer with this project is "How much darker must a color become before it becomes visibly darker than its original?" Artwork becomes darker due to the deposition of fine black carbon particles, or soot. Previous studies had claimed that a 0.2% carbon coverage on a white background would appear visibly soiled to the average human observer. However, the levels determined by our study vary from 2.4% to 12%, depending on the conditions in which the samples are viewed./p>

Art objects are soiled by airborne black elemental carbon particles, otherwise known as soot.

What level of soiling is necessary to produce a visible change in the appearance of an art object?

Approach:

Generate Samples

Soiling was simulated by printing fine dot patterns onto various Munsell QuickColor color standards.

Sections of the soiled color standards were arranged to form the two color tests, the Solid test and the Edge test.

Characterize Samples

Samples were analyzed:

-By optical microscopy to determine actual percent coverage by simulated soot.
-By reflectance spectrophotometry to determine change in color.

Human Testing

30 Human subjects were asked to take both tests:

-Solid Test: Subjects were asked to identify which colored tiles appeared "light" and which appeared "dark".
-Edge Test: Subjects were asked to identify which colored tiles appeared to have a horizontal or vertical line and which appeared blank.

Data Analysis

--For each soiling level for each color, the average accuracy (percent correct) was calculated for both tests.
--Graphs were constructed relating the level of soiling to the percent of responses that were correct.
--An arctangent curve was fit to each graph. The point of inflection of the curve identifies the threshold of detection.

Back to Leon Bellan's personal home page
Back to List of 1999 Intel Science Talent Search Finalists