I have found dates derived from window glass to be very reliable. However,
instead of using the regression formula to calculate one initial
construction date, use it to create a date for each individual fragment of
glass and then create a histogram that accounts for the whole distribution
of window glass fragments recovered from a site.

Briefly, Window glass fragments from each site/structure location were
measured using digital calipers to the nearest thousandth of a centimeter.
These measurements were then processed using Moir’s (1987) formula to obtain
an associated date of manufacture for each individual fragment. These
associated dates were then graphed chronologically by frequency, creating a
histogram. Ideally, the first significant rise of this histogram suggests
the initial construction date of the associated structure. Additional rises
or peaks in the histogram may indicate later additions or remodeling that
took place. This method usually provides a healthy sample size and should
take into consideration long periods of occupation, remodeling episodes, and
possibly the use of scavenged materials.

For a more detailed discussion about window glass dating methods see the
paper I presented at the 4th Annual SCHAC conference entitled, "Window Glass
Dating". This paper is published on-line in PDF format:
http://www.crai-ky.com/projects/reports/windowglass.pdf

Since presenting this paper, I have used this method of dating window glass
for a number of sites in central Kentucky with very good results each time.
I welcome constructive comments and criticisms about this method of dating
window glass.

Grant Day


Grant L. Day, RPA
 Principal Investigator
 Cultural Resource Analysts, Inc.
 Lexington, KY 40508
 Phone: (859) 252-4737
 Fax: (859) 254-3747
 e-mail: [log in to unmask]
 website: www.crai-ky.com




-----Original Message-----
From: HISTORICAL ARCHAEOLOGY on Behalf Of Dylan Stapleton
Sent: Friday, January 24, 2003 4:54 PM
To: [log in to unmask]
Subject: Window glass dating formulas


What is the consensus, if any, on the reliability of window glass dating
regression formulas? . . . .