This is a rather late response to Matthew A. Sterner's posting on setting
up a large data base to record glass manufactures' marks along with the
data on the contexts.  This is an area that has good potential for adding
to our knowledge of the distribution of glass containers and possibly other
questions.  This type of a database could be very simple or rather complex.
The problem is a balance between being too simple and therefore minimal
information return and too complex so that people are not willing to spend
the time to enter the data.

      Let's consider the simplest model of data collection.  Just recording
the makers' trademarks would be the easiest and quickest way to go.
However the data gathered would be of limited value.   A large database of
makers will soon show that the major twentieth-century firms such as
Illinois Glass, Owens Bottle Company, their successor Owens-Illinois, Hazel
Atlas, Ball, Thatcher Manufacturing Company, and Anchor Hocking are found
everywhere.  Recording the distribution of their bottles would be like
plotting snowflakes during a blizzard in Buffalo, New York.  The end
product would probably be a meaningless blobby looking map that would be
more of a record more of where excavations took place rather than market
distribution.

      There are some things to consider in setting up a database to record
marked glass containers.  One is that glass manufacturers produced two
broad categories of bottles.  One type was stock bottles that were sold to
anyone wanting them.  These bottles are generally quite plain.  The second
types were produced by what was called private molds.  Private mold bottles
were made to order of bottlers and manufacturers who wanted a distinct
container for their product.  These bottle molds were either paid for by
the customer or made a part of a package deal to go with a large order of
bottles.  Private mold ware is associated with one company, and the
distribution of that product would probably say a lot more about regional
and national markets than just recording the glassmakers' marks.

      Private molds can be broken down into two major groups.  Most private
molds were used once and not brought back for refilling.  The distribution
of these bottles will be a pretty good record of the market for the
products they held through time.  The second group of bottles includes
those that came with a deposit and were returned to get the deposit back so
that they could be refilled.  For products from small companies with a
local market such as dairies or small breweries, the distribution
information will be quite useful.  For products with a national market that
was put up by regional bottling plants the questions related to
distribution are not as easy to deal with.  Coca Cola is the prime example
here.  Coca Cola has bottling  plants in every state, and in several states
they had more than one bottling plant.  According to Glass Containers 1964,
published by the Glass Container Manufacturers Institute, the average soft
drink bottle made 22 trips for refilling before being broken or discarded
(page 7).  The same report stated that the average milk bottle made 30
trips and the average beer bottle made 26 trips before breaking or being
discarded.   In the late 1950s when I was a bag boy in a small local
grocery store, one of our jobs was to sort pop bottles for pickup by the
various venders.   We had a running game of who could find the Coke bottle
from the farthest location.  We commonly had bottles from California,
Texas, and other distant locations.  Some collectors of Coke bottles
attempt to acquire bottles from all of the states.  The Coca Cola Company
is aware of this and their special non-returnable small classic shape Coke
bottles brought out at Christmas time are now marked with different states
on the bottom which have nothing to do with where the bottles were made.
Last year I purchased a six- pack of these disposable bottles, and the
bottles had six different states on them.  This seems to b a rather crass
appeal to Coke bottle collectors and putting information on the bottle that
has no relationship as to where the Coke was bottled.  Clearly, using the
information on a Coke bottle for studying regional and national markets is
fraught with difficulties and problems.  In summary, private mold bottles
for identifiable products with non-reused bottles have a great potential
for looking at regional and national markets.

      Beyond private and stock molds there are other types of information
available from markings on bottles that would be useful to gather into a
database because they have good potential to provide chronological
information.  Machine-made bottles are clearly going to be the clock for
dating twentieth century sites.  This is an area where much can be done to
provide tighter dates than are available for any other time period we deal
with.  Here is a list of some of the possible things that would be worth
recording.

Bottle mold numbers

      Almost all glass factories used bottle mold numbers as a way of
keeping track of their inventories and to help customers order bottles.
Surviving glassmakers' catalogs provide commonly provide the size and name
and an illustration of their bottles along with the mold numbers.
Compiling a list of the mold numbers along with the names of the bottle
type and size would provide an excellent tool for identification of bottles
and their makers.  The problem is that different glass factories use the
same numbers. This is especially true for three digit numbers.  Thus if you
have three bottles with only the number "50" embossed on their base and
they are quite different in shape and size, it is impossible to know who
made the bottle.  However, the systematic gathering of mold number with
descriptions from catalogs could increase our ability to identify and date
bottles.  Bottle mold number information has already been compiled for
Canadian glass companies from company catalogs.  (Miller and Jorgensen
1986).

Date codes

      Several of the glass companies used date codes that are generally
described in Toulouse.  In some cases such as Owens-Illinois Glass Company
the date code is a one-digit number to the right of the trademark.  Thus if
you have a "3" in that position the bottle could be from 1933, 1943 or
1953.  The date code can often be used with other characteristics of the
bottle to begin to eliminate some of the numbers.  For example, if the "3"
is on a cork stopped bottle, it probably predates WWII and thus would have
been made in 1933.  Date code numbers should be included in any database
recording system.

Plant codes or locations

      Again, the Owens-Illinois Glass Company had a number to the left of
its trademark.  Toulouse records the locations along with beginning and end
dates for many of these plants (Toulouse 1971:395).  The plant number in
combination with the date code can sometimes help tighten the date.  All
numbers on the base of trade marked bottles should be recorded because we
may find out what they mean by further research.

Technology of production

      Some glass manufacturers, such as Illinois Glass Company, made the
transition from mouth-blown bottles to machine-made bottles.  Recording the
technology used to make the bottle could go a long ways to improving our
ability to date bottles.  Here it might be best to have a checklist for the
types i.e. mouth-blown, machine-made, Owens machine-made, and suction scar.
The Illinois Glass Company changed the marks it used and perhaps that
change has a relationship to the technology.  We will not know that unless
the data are collected.  If that relationship can be established, then a
small sherd with part of the I-Diamond mark could be identified as to
whether it was mouth-blown or machine made based on the information
recorded in the data base.  Dates for makers' marks can be sometimes
tightened up by the technology used.  The dates for when different
companies gained access to the Owens bottle-blowing machine, for example,
are well documented (Miller and McNichol 2002).

Glass color

      The movement toward colorless glass is an area where there is good
potential for improving our ability to date bottles.  One area that has
received a lot of attention is that of solarized glass.  This glass was
made colorless by the addition of manganese to the glass batch to overcome
the slight green color from iron in the glass (Miller and Pacey
1985:44-45).  Manganese works well in glass melted in crucibles where it
was easy to maintain an oxidizing atmosphere.  It did not work very well in
tank furnaces where it was difficult to maintain an oxidizing atmosphere.
By 1917 half of the bottles being produced in the United States were being
made on the Owens Bottle Blowing Machine that was dependent on tank
furnaces (Miller and Sullivan 2000: 166-167).  I have yet to see a
solarized Owens made bottle.  When one finds a machine-made bottle that is
solarized, it probably means that it was produced on a semi-automatic
bottle-blowing machine that was hand fed from a crucible.

      Early machine-made bottles are often light green in color and it
appears that these light green bottles were on their way to being replaced
by colorless bottles during the 1920s.  In addition to the light green
bottles, the other dominant colors for machine-made bottles were amber,
green, and cobalt blue.  This seems to be the colors that the glassmakers
worked and I can not see any need to go beyond these basic color listings
which could be in a simple check list form in a data base for the following
colors:
Colorless, pale green, green, amber, blue, and other.  Some products such
as the Coca Cola bottle and the Ball canning jar had their own special
color that does not fit into this above progression and it would be silly
to force these bottles into the color scheme.

Summary

      Already, I have suggested several things that would lengthen the
database and perhaps discourage people from entering their data.  However,
these traits should be considered in terms of what they might offer in
terms of information return, both in chronology and distribution system.  I
would find it difficult to get excited about a system that limited itself
to distribution systems.

George L. Miller
URS Corporation
561 Cedar Lane
Florence, New Jersey  08518

Miller, George L. and Elizabeth A. Jorgensen
      1986  Some Notes on Bottle Mould Numbers from the Dominion Glass
            Company and its Predecessors. Parks Canada, Ottawa. Available
            from Research Publications, Parks Canada, 1600 Liverpool Court,
            Ottawa, Ontario  Canada  K1A 1G2  a free publication.
Miller, George L. and Tony McNichol
      2002  Dates for Suction Scarred Bottoms: Chronological changes in
            Owens Machine-made Bottles.  Paper given at the 2002 SHA
            meetings, Mobile, Alabama.  Email senior author for a copy.
Miller, George L. and Antony Pacey
      1985  The Impact of Mechanization in the Glass Container Industry:
The Dominion Glass Company of             Montreal, a Case Study.
Historical Archaeology 19(1)38-50.
Miller, George L. and Catherine Sullivan
      2000  Machine Made Glass Containers and the End of Production for
Mouth Blown Bottles. Pages                161-174 in Approaches to Material
Culture Research for Historical Archaeologists.  David R.         Brauner,
compiler.  Society for Historical Archaeology.
Toulouse, Julian Harrison
      1971  Bottle Makers and Their Marks.  2001 reprint by Blackburn
            Press, Caldwell, New Jersey.