Analysis shatters cathedral glass myth
by C. Wu
A new study debunks the persistent belief that stained glass windows in medieval cathedrals are thicker at the bottom because the glass flows slowly downward like a very viscous liquid.
Edgar Dutra Zanotto of the Federal University of Sao Carlos in Brazil calculated the time needed for viscous flow to change the thickness of different types of glass by a noticeable amount. Cathedral glass would require a period "well beyond the age of the universe," he says.
Suffice it to say that the glass could not have thickened since the 12th century. Zanotto reports his finding in the May American Journal of Physics.
The study demonstrates dramatically what many scientists had reasoned earlier. "You would have to bring normal glass to 350° Celsius in order to begin to see changes," says William C. LaCourse, assistant director of the NSF Industry-University Center for Glass Research at Alfred (N.Y.) University.
Viscosity depends on the chemical composition of the glass. Even germanium oxide glass, which flows more easily than other types, would take 1032 years to sag, Zanotto calculates. Medieval stained glass contains impurities that could lower the viscosity and speed the flow, but even a significant reduction wouldn't alter the conclusion, he remarks, since the age of the universe is only 1010 years.
The difference in thickness sometimes observed in antique windows probably results from glass manufacturing methods, says LaCourse. Until the 19th century, the only way to make window glass was to blow molten glass into a large globe then flatten it into a disk. Whirling the disk introduced ripples and thickened the edges. For structural stability, it would make sense to install those thick portions in the bottom of the pane, he says.
Later glass was drawn into sheets by pulling it from the melt on a rod, a method that made windows more uniform. Today, most window glass is made by floating liquid glass on molten tin. This process, developed about 30 years ago, makes the surface extremely flat.
The origins of the stained glass myth are unclear, but the confusion probably arose from a misunderstanding of the amorphous atomic structure of glass, in which atoms do not assume a fixed crystal structure. "The structure of the liquid and the structure of the [solid] glass are very similar," says LaCourse, "but thermodynamically they are not the same."
Glass does not have a precise freezing point; rather, it has what's known as a glass transition temperature, typically a few hundred degrees Celsius. Cooled below this temperature, liquid glass retains its amorphous structure yet takes on the physical properties of a solid rather than a supercooled liquid.
"At first, I thought that the [sagging window idea] was a Brazilian myth," Zanotto wrote, but he soon learned that people all over the world share the belief. Repeated in reference books, in science classes, and recently over the Internet, the idea has been repeatedly pulled out to explain ripply windows in old houses. "For the layperson, it makes a lot of sense," says LaCourse.
In 1989, Robert C. Plumb of Worcester (Mass.) Polytechnic Institute suggested in the Journal of Chemical Education that definitive proof might require an instruction book written in the Middle Ages advising glaziers to install glass panes with the thick end at the bottom. Now if only such a handbook could be found.
From Science News, Vol. 153, No. 22, May 30, 1998, p. 341.
Copyright Ó 1998 by Science Service.
Zanotto, E.D. 1998. Do cathedral glasses flow? American Journal of Physics 66(May):392.
Plumb, R.C. 1989. Antique windowpanes and the flow of supercooled liquids. Journal of Chemical Education 66:994.
Additional information about antique windowpanes and glass flow can be found at http://www.ualberta.ca/~bderksen/florin.html and http://www.ualberta.ca/~bderksen/windowpane.html .
William C. LaCourse
New York State College of Ceramics at Alfred University
NSF Industry-University Center for Glass Research
2 Pine Street
Alfred, NY 14802
Robert C. Plumb
Worcester Polytechnic Institute
Worcester, MA 01609
Edgar D. Zanotto
Federal University of São Carlos
Department of Materials Engineering
13565-905 São Carlos-SP
copyright 1998 ScienceService