Most of the literature on containers has appeared in the last ten years or so and is the direct result of the growing popularity of bottle collecting, particularly in the United States.  The literature tends to concentrate on collectors' interests in bottles, such as the identification and price, and to ignore or be uninformed on the historical aspect (Jones 1975). 

     With unusual exceptions, the dating of glass containers cannot be restricted to a single year (even if the year of manufacture is embossed on the bottle), but rather encompasses a number of years or a date range.  These ranges extend from the actual date of manufacture (beginning date) to the date of most recent use before discard (end date).  Each bottle, therefore, has two dates which often must be determined separately. 

     Bottle dating can be relatively simplistic or extremely complex depending on several factors, including the experience of the researcher, the goals of the project, and the availability of background information.  The most simplistic design includes the most probable date of container use rather than the extremes of possible usage.  To illustrate the difference between possible and probable dates, we shall examine a neck/finish fragment excavated at San Elizario, Texas (Lockhart & Olszewski 1974:42-44).  The finish consists of two applied rings, the upper one much longer in vertical measurement than the lower, commonly known as a brandy finish.  The neck, itself, is slightly swollen, a condition generally indicating a container manufactured as a beer bottle.  The color is a dark amber with mold marks extending from the break at the neck/shoulder junction to a point about halfway to the finish. 

     A probable beginning date would be 1881, the year the railroads first arrived in El Paso, and, coincidentally, the time when large quantities of bottled goods (including beer) first become evident in the archaeological record of the area (Lockhart 1995:170-171).  The probable end date is about 1895 to 1900, the approximate time that crown finishes replaced all other finishes on beer and soft drink bottles (see below).  The possible dates extend beyond the probable in both directions.  The early date must be altered to include the earliest possible date of the use of the two-piece mold technique (1750 for small bottles, although ca. 1845 is much more likely for larger bottles such as this one) and even the earlier three-piece mold technique (1810) because either manufacturing method could have produced mold lines that end halfway up the neck.  The swelled neck indicates a greater likelihood that the manufacture was later, but the possibility of a swollen neck on an earlier bottle cannot be ruled out.  Similarly, the end date must be extended for an indefinite period, probably not exceeding 1920, because every individual glass house did not immediately switch to crown caps when they became available.  Thus the possible date (1810-1920) is ninety-one years longer than the probable date (1881-1900).  The greater the quantity of datable artifacts that are found on a given site, the greater the confidence in the probable date range rather than the possible.  A single artifact (such as our neck/finish fragment) must be given the widest possible date range, while the same artifact found in quantity (225 similar finishes were excavated at San Elizario) may more comfortably be dated within the probable range. 

     Bottles may be dated by a variety of methods that will each produce a date range that is not necessarily the same as the date range indicated by the other methods.  These methods include examining the morphology (shape) of the container, glass color, manufacturing techniques, manufacturer's marks, retail labeling, embossed dates, and patent or copyright dates.  Although the use of a single method will suffice in some cases, a combination of methods will generally produce the smallest, most accurate range of probable or possible dates. 

 although classification by colour is simple to do, the end result is of little value for the following reasons:  color does not have a direct relation with glass type. . . colour is not related to the technology of glass object production. . . colour is only weakly related to the function of the object. . . . Given these factors there is little justification for using colour as a means of classification. . . .There is a very broad chronology of popularity of various colours over time; however that chronology cannot be applied to individual glass objects with any significant level of meaning (Jones & Sullivan 1989:12). 

Jones, however, used color/glass composition in conjunction with other analytical tools on at least one occasion but still warned that "glass color alone cannot be used to determine glass composition" (Jones 1983:71).  Others, such as Fike (1987), Newman (1970), and Staski (1984) accept the color of glass for at least limited use.  While all these authors have valid points in their critiques of color usage as a diagnostic tool, the concept of color can be an acceptable dating technique if carefully applied. 

     At least two colors are generally accepted as having validity in dating glass artifacts.  The first of these, black glass, is really an intense dark green or purple glass that is almost completely opaque, displaying very little translucency.  Its primary use was in liquor or wine bottles in the seventeenth through nineteenth centuries.  Although authors have differences of opinion on exactly how to date black glass, it can be generally stated that it occurs prior to 1880 (Fike 1987:13; Jones & Sullivan 1989:14; Newman 1970:74). 

     Solarized purple or amethyst glass forms the second color classification that is generally accepted as a valid color in artifact dating.  The amethyst or purple color is derived from the prolonged solar exposure of colorless glass that was manufactured with manganese as a decolorant.  Iron impurities within normal glass composition create an amber to light blue to aquamarine or blue-green color depending on the type and amount of iron in the mixture.  Manganese reacts with the iron to produce a colorless effect in the final glass product. Prolonged exposure to the ultraviolet rays of the sun, however, causes a chemical change in manganese that creates varying shades of purple or amethyst depending on the concentration of manganese in the glass mixture.  Concentration of manganese also affects the necessary duration of solar exposure required to produce the color change.  Because of varied concentration, required duration of exposure to natural sunlight can be as little as two weeks or over a year before a color change is introduced. 

     Although manganese was used in glass (mostly to create purple color, rather than as a decolorant) as early as the Roman Empire, its use in decoloring glass in the United States is fairly recent, with 1880-1914 as the often-accepted dates of usage set by Kendrick (1963:54-55), although recent historical/archaeological research revises these to ca. 1890-1920 (Lockhart 2000).  These, however, are probable dates, not possible ones.  According to one source, possible beginning dates begin as early as 1810, although colorless glass was unusual at that time (Zimmerman 1964:8).  Possible end dates also extend farther toward the present, with manganese used as a decolorant by small glass houses as late as the mid-1930s (Sharp 1933:763).  Thus, quantities of solarized purple or amethyst glass may be dated between 1890 and 1920 with a good level of confidence, but the use of probable dates in assessing a single artifact must be approached with caution. 

     A second problem related to manganese decolored glass arises for the archaeologist.  Because solarization requires exposure to the sun, large quantities of excavated manganese-bearing colorless glass may not have absorbed the necessary amount of ultraviolet radiation to produce the color change.  As bottle collectors have been aware for many years, solarization may be achieved in a much shorter period by using artificial sources of ultraviolet radiation such as sunlamps or germicidal lights.  Glass that is suspected of having a deposition prior to 1920 can be tested relatively cheaply by means of such a light enclosed in a wooden box constructed with aluminum foil lining to reflect and intensify the ultraviolet light (Cf. Lockhart 2000). 

     Colorless glass itself provides a beginning date when applied to aggregates.  The manufacture of colorless glass containers came into prominence around 1890, although some colorless glass was used prior to that time.  Because colorless glass became so popular, the manufacture of the naturally tinted light blues and aquas began to decline until the use of unaltered iron-impure glass had mostly declined by about 1910, especially in the manufacture of soda and beer bottles.  With no other evidence, however, the 1910 date must be used with caution because it is a probable, not possible, time frame.  In addition, glass in the early Royal Crown Cola bottles and the Georgia Green Coca-Cola bottles closely resembles the older aqua-colored glass. 

     A brief statement must be made about the use of the word "colorless."  Popular literature, as well as that of historical archaeology, has traditionally used the word "clear" in describing glass that lacks noticeable color.  The word "clear," however, is ambiguous, meaning among other things, transparent, unmistakable, clean, pure, certain, or unencumbered by debts.  In describing glass, it is generally used to mean transparent and colorless.  Other colors, however, can be transparent--especially greens and blues.  Colorless, on the other hand, clearly means without color (Cf. Jones & Sullivan 1989:13). 

     Although numerous colors were used prior to 1900, the most common ones were after 1880 were aquamarine (common green), solarized amethyst, and colorless.  Other glass colors were rarely used in the later manufacture of soft-drink bottles, although amber containers enjoyed a brief popularity in the early twentieth century.  Like many other colors, amber is best used as a dating tool in conjunction with one or more other physical attributes of a container and/or one or more other dating techniques.  Amber soda bottles provide a good example.  If a container is the right size and shape to be a soft-drink bottle and, especially, if it is embossed with a clear indicator, such as "BOTTLING WORKS" (a term generally discontinued by the mid-1930s), then the amber color will suggest a date range in the first two decades of the twentieth century.  However, not every player in the game used the same rule book.  The makers of Orange Crush, for example, continued to produce their product in amber bottles after 1935.  Aside from the cereal beverages (near-beers) of the Prohibition era (usually paper labeled), only Whiz and Orange Crush have thus far been discovered as users of amber colored soda bottles in El Paso. 

     Greens are equally undatable, except for the above-mentioned Georgia Green of Coca-Cola that first appeared in the company's 1915-patented "hobble-skirt" bottle (not actually in use until 1916 and then, not in all franchises until ca. 1920) and the equally distinguishable Forest Green that was popularized by Seven-Up around 1928.  Similar colors have been used by other bottlers, but that use generally occurs after the Coca-Cola and Seven-Up dates.  In older bottles (i.e. pre-1930), a frequently used color was the aquamarine tint usually created by the iron impurities inherent in almost all glass mixtures.  Because these impurities were most easily dispensed with by ignoring them, the resultant color was called "common green" as early as the seventeenth century (Harrington 1952:33-34), a practice I will emulate in this volume. 

     Although they rarely occur in soft-drink bottles, probable dates are also available for cobalt blue and opaque glass (generally called milk glass).  Cobalt blue, although probably the earliest glass color (Dayton 1993:1-2, 27-31), saw general use in the United States between the 1890s and the 1960s.  Opaque glass is often called milk glass because the commonly-used white color resembles the hue and consistency of cow's milk.  Opaque glass, however, is occasionally found in other colors, notably pastel blues and greens.  Like cobalt blue glass, opaque glass was generally used in containers between the 1890s and 1960s (Fike 1987:13). 

Figure 2-1 - Round-based Rossi's Belfast bottle

     To insure a good seal, bottles with Hutchinson finishes were packed upside down in special wooden cases.  Holes the appropriate size for the finishes were drilled in the bottom of the cases and the necks were inserted to maintain the upside down position.  Because of this method of packaging, most of the heavy wear (i.e. chipping) appears on the base and heel areas of the bottles (personal communication, Ken Malone, 9/30/2000).

     One of the jobs of the route salesmen (or anyone else collecting bottles for the company) was to insure that all bottles in the cases belonged to the bottler (no "foreign" bottles).  To assist in the process, many bottlers had their names embossed on the bases of their containers.  A route salesman could then easily tell if any foreign bottles had been included.  The practice continued on crown finish bottles, even though the need no longer existed.  Some El Paso bottlers continued using embossed-base bottles well into the 1930s, although the practice had been generally eliminated by about 1910.

     Swing stoppers, such as the Lightning Stopper, manufactured by Henry W. Putnam were rarely used on soft-drink bottles.  With stoppers of this variety, the seal was affected by a disk or plunger arrangement that sat atop or just within the bottle lip and was held in place by a set of stiff, spring-loaded wires.  When the controlling wire was released, the disk or plunger swung away from the opening to allow the contained liquid to pour forth into a glass or thirsty gullet.  Like the Hutchinson stopper, the Lightning stopper could be resealed by reversing the process.  Such stoppers were first patented in 1878 and used into the twentieth century (Paul & Parmalee 1973:14).  No known El Paso soft-drink bottlers used the Lightning or other similar stoppers. 

     Numerous other varieties of stoppers were marketed at various times but were never as successful as the ones discussed above.  Codd or ball-type stoppers may have occasionally found their way to the El Paso area but were never used by local bottlers.  Named for Hiram Codd, these ball-type stoppers used a glass or hard rubber ball located inside the bottle (usually in the neck) to provide the seal.  The ball was held in place by the pressure from the carbonation in the liquid (Paul & Parmalee 1973:10-12).  William Painter's Baltimore Loop Seal consisted of a hard rubber disk that sat in a specially molded groove inside the throat of the bottle.  Initially, the bottle was opened by  pushing the seal into the container or prying it out with a special tool.  Eventually Painter added the loop which protruded from the top of the seal and aided in its removal by prying (Lief 1965:16-17).  Although the Codd stopper was popular in England, neither of the two achieved much success in the United States.  Occasional bottles using ball stoppers may have found their way into the El Paso area, but I have been unable to locate evidence that any bottles utilizing Baltimore Loop Seals were ever used in the city.  Other inventions and variations of the late nineteenth century were even less popular. 

     The final method of closure used on returnable bottles by soft-drink bottlers was the Crown Closure.  Invented by William Painter, the device went through a series of modifications prior to the final patent issued on February 2, 1892.  The simple device consisted of a rolled steel cap that was crimped over a special finish to create a crown-like effect.  A cork disk in the center of the crown cap (later replaced by plastic) provided the seal.  Although the new closure was simple and effective (so much so that it is still in use today), acceptance by bottlers and the public was slow--possibly fueled in part by the depression of 1893. As an experiment, a brewer shipped a cargo of beer closed with crown caps to South America and back with the bottles unopened.  Despite time, heat, and the rough treatment in shipping, the beer was unflawed (Woodroof & Phillips 1974:222). Even though Paul and Parmalee (1973:14-20) suggest that less than twenty-five percent of American bottlers were using the crown cap by 1905, El Paso bottlers appear to have accepted the concept early.  There is no record of when Houck & Dieter switched from Hutchinson stoppers, but the majority of their containers that survived in local collections used crown caps.  All known bottles from Henry Pfaff (late 1898-1907) and the last bottle style used by his predecessor (also bearing the initils of Henry Pfaff), R. F. Johnson & Co. (1895-1898) used crown caps as did the earliest bottles from Purity Bottling and Manufacturing Company (1906-1912), Magnolia Bottling Company (1908-1914), and Woodlawn Bottling Company/Pepsi Cola Bottling Company (1909-present).  On the other extreme, the remaining R. F. Johnson (1891-1895) and El Paso Bottling Company (1892-1895) bottles known to exist used Hutchinson stoppers.  The obvious inference is that crown caps had became popular in El Paso by later half of the 1890 decade (1895-1900). 

     The early crown finishes were formed by hand after the bottle was blown into a two-piece mold by the gaffer (master blower).  Although the inside and outside diametric measurements of these finishes are remarkably consistent (2.6 cm. outside lip diameter; 1.6 cm. inside), the length of the neck can vary considerably.  On Henry Pfaff brand bottles, height from base to neck varied from 20.4-21.3 cm., a range of 0.9 cm.  Interestingly, the same bottles were fairly consistent in average diameter (6.2-6.3 cm.), although the bottles were not perfectly circular in cross-section.  Individual Pfaff bottles had diameters that varied from 6.1-6.3 cm. depending on the axes along which the measurement was taken.  This lack of perfect roundness was relatively consistent well into the 20th century, and many bottles into the 1950s were less than perfectly cylindrical. 

     Crown caps were locally available from Crown Cork & Seal Company from 1920 to 1970, although they were shipped in from out of state sources prior to that time.  Crown Cork & Seal was first located at 413 E. Boulevard (changed to Yandell Blvd. the following year) but moved to 1601 Magoffin Avenue in 1938 and finally to 1701 Bassett Avenue in 1962 (EPCD 1938-1970; EPTD 1920-1970). 

     A Semi-automatic glass blowing machine was patented for manufacturing wide-mouth containers as early as 1881, but application to small-mouth bottles did not follow until 1886. Actual commercial production, however, did not begin for another three years.  The new technology was slow to be applied because of the powerful glass-blowers unions.  Michael Owens patented the first fully-automatic bottle machine in 1903 and began the slow task of completely revamping the bottle manufacturing industry.  Although the fully automatic machinery vastly improved both the speed and quality of production, it, too, was slow to be accepted.  The new machine, like the semi-automatic process before it, was initially manufactured for the production of wide-mouth containers, and it was not until 1907 that the process was perfected for use in the making of soft drink bottles.  By 1917, only about half of all bottles in the United States were produced by automatic bottling machines.  The most important attribute of machine-produced bottles was the standardization of sizes; a single closure could now fit any bottle that emerged from the manufacturing process (Miller & Sullivan 1984:35-39). 

     Recognition between two-piece mold and automatic machine techniques is relatively easy.  The seams where the two parts of the mold joined together left mold lines on both the base and sides of the finished bottle.  Although basal mold lines in the two-piece mold technique could encircle the circumference of the base or extend across it, two distinct lines always extended from base to a point above the body/shoulder junction.  Because the top of the bottle was completed by hand (hence the use of the name "finish" for the upper extremity of the container), the mold lines were erased as the gaffer applied the finish.  Although earlier literature suggested that the distance between the end of the mould line and the lip of bottle could be used to indicate a relative date (Kendrick 1963:45-47), the length of the mold line is more indicative of the skill of the individual gaffer than any industrial changes.  The suggestion that a mold line ending close to the lip of the bottle indicates a semi-automatic bottle machine is incorrect (Newman 1970:72). 

     Both semi-automatic and fully-automatic bottle machines reversed the traditional process used by the earlier techniques and began by forming the finish.  As a result, the mold lines on the container either extend to the lip or join other mold lines that encircle the finish at various points.  Wandering "ghost" seams are another characteristic of machine-made bottles.  These seams may run parallel to the more distinct mold lines or may be curved.  Finally, small, circular valve marks on the base indicate a machine manufacture, as do "ghost" seams on the base.  A final basal characteristic is the distinctive ring that identifies an Owens Automatic Bottle Machine.  The presence of the Owens ring may also be used as a dating device.  Miller and Sullivan describe the Owens ring as "a distinctive circular mark with ‘feathery' edges, caused by the shears that cut off the bog of glass in the suction machines.  An Owens scar is usually off-center and may sometimes even extend onto the heel.  It dates from 1904 until at least 1969" (1984:93)..  There is no readily available method to easily distinguish between semi-automatic and fully-automatic bottle machines. 

     The process called the two-piece mold technique actually used a three-piece mold by the time soft drink bottles were being filled in El Paso.  An actual two-piece mold leaves a single line across the base of the bottle where the two halves joined.  In the later processes, another mold piece on the base was shaped either as a cup or a post.  The cup left a mold mark on the heel of the bottle; the post left a circular mark on the base.  The use of a plate (see below) required a fourth piece. 

     Dating the changes becomes more complex than explaining them.  Use of the two-piece mold began around 1845 (1750 in England for small bottles), although El Paso soft-drink bottles all date well after that period.  The termination of the process is less well defined.  Although hand blown containers accounted for only about fifty percent of the total United States bottle production by 1917, the automatic process did not lend itself well to small orders.  Small orders, especially those demanding oddly-shaped bottles, continued to be filled by the two-piece mold technique until at least the mid-1930s.  Therefore a possible date would have to extend to 1935, although a more probable date, especially for the production of soft-drink bottles, would be about 1915.  The possible beginning date for machine-made bottles is 1889, although a more probable date, in most cases, would be around 1913 (Newman 1970:70-71; cf. Miller & Sullivan 1984:88-90, 93-94 for possible dates).   Paul & Parmalee (1973:25) maintain that "the automatic bottling machine was not operational for the production of pop bottles for some time, and it was not until about 1915 that most bottles were so manufactured."  The manufacture of machine-made glass containers, of course, continues today.  Although the manufacture of soft drink bottles by machine was first introduced in 1907, such bottles were apparently not used in El Paso until much later.  Evidence from El Paso soft-drink bottles supports the probable dates, especially the end date of 1915.  Almost all (at least one exception) known Empire Bottling Works containers (1912-1922) are machine-made.  Bottles from the Purity Bottling and Manufacturing Company (1906-1912) and Houck & Dieter (1881-1912), who merged to form Empire, are manufactured by the two-piece mold process.  Magnolia Bottling Company changed to machine-made bottles sometime after 1911 but before 1920, when the company began bottling Cola-Cola.  Because both Magnolia and Empire continued using two-piece mold bottles after 1911 and 1912 respectively, and because such bottles are scarce, the date of the change was probably not long after 1912.  A reasonable estimate for the general switch of El Paso bottlers to machine-made bottles, therefore, would be about 1915. 

     Many soft drink bottles during the period from about 1865 (1881 in El Paso) to the early 1920s were manufactured using  plate-molds (frequently called slug plates in collectors' literature).  An adaptation of the two-piece mold process, the "plate" was a circular, oval, or tombstone-shaped insert set into the front piece of the mold.  The debossed name of the local bottler was engraved on the plate which could be inserted in the mold form at much lower cost than retooling an entire mold for each bottler.  The plate outlines often left a visible mold line (frequently quite distinct) on the surface of the bottle (Jones 1983:173; Paul & Parmalee 1973:21).  Most of the pre-1920 bottles used in El Paso displayed obvious plate-mold markings.  New information could also be added to plate molds (Pollard 1993:344).  For example, Woodlawn Bottling Co. had THIS BOTTLE/TO BE RETURNED added between WOODLAWN BOTTLING/CO. and EL PASO, TEX. on one of their plate-mold bottles. 

     Elaborately embossed "proprietary" bottles generally replaced the plate-mold styles in the early 1920s.  These bottles often included embossed rings, vertical panels and ridges, oblique lines, stars, "orange-peel" surfaces, or line drawings in addition to the national franchiser's name, logo, and advertising information.  Although some of these contained the bottler's location embossed on the base, few of the national brands identified the local bottler by name.  Proprietary styles persisted until at least the mid-1930s to early 1940s when they were replaced in general use by Applied Color Labeling (ACL).  The 1915 "hobble-skirt" Coca-Cola bottles were an early example of proprietary bottles.  Many companies used paper labels in conjunction with embossing, especially during the pre-1920 period, but the practice extended until the beginning of ACL labeling (and sometimes later).  Although paper had the advantage of offering a stage for complex advertising, labels had to be replaced with every washing at an additional expense.  In addition, the glue used to apply paper labels was prone to dissolve with prolonged immersion in water.  The early method of display for cold soft drinks was generally a bucket, barrel, or metal box filled with ice.  As the ice melted, it formed a reservoir of cold water providing an ideal environment for unintentional label removal.  Resulting dissatisfaction led to the rapid acceptance of baked enamel labeling, which generally replaced paper labels in the mid-1930s (Paul & Parmalee 1973:25-28). 

     Paper labels also adorned a unique type of soft drink bottle that was popular during the prohibition era (1920-1933--1919 in Texas).  These were the near-beers that were generally bottled in beer bottles, often by beer companies that were trying to remain solvent through the alcohol-free period.  Such bottles usually become archaeologically "invisible" because most paper labels fail to survive the rigors of depositional processes, and the bottles themselves are indistinguishable from normal beer bottles.  El Paso bottlers distributed such brands as Barlow, Bock, Bone-Dry, Bravo, Falstaff Dublin-Style Cereal Beverage, Falstaff Pale, Goldcrest, Golden Glow, Graino, and NIB (Non-Intoxicating Beverage).  The names of some brands (e.g. Bone-Dry and NIB) obviously played on the theme of Prohibition. 

     Archaeologists are frequently unconcerned with paper-label containers because of their poor preservation characteristics; however, paper-label bottles do survive under some conditions.  Occasionally, the paper labels themselves survive.  Wilson (1981:3-6, 24, 54) excavated several paper-label-bearing containers from Fort Union, New Mexico, and Fort Laramie, Wyoming, including beer, bitters, and medicine types, some as old as the 1860s or 1870s.  Berberich (1973:16-17) found 60 wine bottles deposited by an 1875 or 1876 flood, many of which retained their labels.  In some cases the label disintegrates but the dye remains.  The author excavated three beer bottle fragments in San Elizario, Texas, that retained the full impression of the dye, although the paper was long decomposed.  The remaining dye was quite delicate and crumbled from the glass at a touch (Lockhart & Olszewski 1994:40-41). 

     The baked enamel labeling (or ACL), was initially used on soda bottles in 1934 but was not common throughout the industry until the early to mid-1940s. From its inception, the process became extremely popular with the soda bottling industry; enameled labels still appear on bottles today.  The process allows for minute lettering that remains legible, an advantage unavailable with the less sophisticated method of embossing.  The process was inexpensive and durable.  The enamel, itself, was "a paste-like glass material consisting of finely ground lead, borosilicate glass and mineral pigments, [which] was forced through a screen mesh directly onto the bottles (sic) surface" (Sweeney 1995:6).  Embossing, however, frequently accompanied ACL labels, usually in the form of designs on the body or manufacturer's markings on the bottle base.  Some of the earliest ACL users in El Paso tried bichromatic or even polychromatic (usually tri-color) designs, although many quickly retreated to monochrome labels because of cost effectiveness.  However, many national brands used bichrome labeling.  Four-color polychromatic labels only appear on commemorative bottles, and most of them are relatively recent. 

     Because each color was applied separately, one problem involved with bichromatic or polychromatic enameled labels was the correct placement of the second (or third) color.  The simple solution was usually an embossed or debossed "dimple" at the container's heel that allowed the ACL machinery to align the bottle identically during each color application.  Dimples were square, round, and rectangular or, occasionally, even long slots tapering gradually deeper into the heel.  Some were very deep protrusions (or intrusions) while others were quite shallow.  Although these dimples aligned the separate color applications, they were sometimes not synchronized with embossed messages on the bottle's heel.  Frequently, embossed contents information appeared on the heel, and this embossing was often as much as ninety degrees out of alignment with the enameled labels.  Good examples of dimples appear on photographs of the Chocolate Soldier bottle and the Mr. Cola, Jr. bottle. 

     Local bottler information returned with the inception of the ACL process.  Because additional information could be easily added, local bottlers identified themselves on many national brands, including Seven-Up, Royal Crown Cola, Pepsi-Cola, Barq's, Orange Crush, and Canada Dry.  Others did not.  Nesbitt's identified the local bottler on the cap.  During the mid- to late 1950s (ocasionally early 1960s), local bottler information again disappeared from labels, leaving only the national franchiser's name and (frequently) address. 

     Although the vast majority of soft drink bottlers complied with the industry standards, a few retained their older designs.  The notably conservative Coca-Cola Company continued to use embossed bottles until the general changeover to plastic containers, although the company finally added the word, Coke, in ACL lettering in 1958 (Munsey 1972:63).  Dr Pepper also held out against the ACL process but based their design on different reasoning.  One of the problems with embossed bottles had to do with the washing and filling process.  During these operations, the bottles were frequently rubbed against each other causing an abrasive effect that defaced the embossing and creating an unappealing "ground" effect at the surface of each embossed letter or decoration.  Dr Pepper's answer to the problem was to deboss their clock logo and lettering.  Although they changed designs in 1950, Dr Pepper continued to use the debossed bottle from about 1930 to 1954 when an ACL transition bottle appeared briefly to be replaced by a container completely labeled with baked enamel lettering in June of 1955 (personal communication from Mildred G. Walker, Curator of Collections, Dr Pepper Museum, Waco, Texas). 

     The Cliquot Club Company was the first experimenter in the use of screw-top cans for soft-drinks in 1938, but their ginger ale ate through the containers terminating the venture.  Pepsi Cola followed in 1950, but exploding cans caused the abandonment of their project.  Cantrell and Cochran (C & C) were more successful in 1953, sparking an industry-wide use of cans as soft-drink containers (Rock 1987:60; for information on the development of cans for the beverage industry, see Woodroof & Phillips, 200-202, 207-211).  The use of pull-tabs began in 1962 and lasted until 1974, causing the demise of the venerable "church key" can opener that had traditionally been used to gain access to canned liquids.  Ring-tabs were introduced in 1965 and continued to be used along with the slightly older pull-tabs until both were outmoded by non-removable pull tops in 1974.  Aluminum cans were first marketed in 1959 and quickly became the industry standard, although some bottlers continued to prefer the rolled steel "tin" variety. 

     In 1927, Coca-Cola had introduced the first "one-way" bottles for use on ocean liners, but non-returnable glass bottles did not become popular until after 1948 (Munsey 1972:72).  Plastic bottles were first used in the late 1950s, but were not popular with bottlers until around 1970.  Once under way, non-returnable glass bottles, cans, and plastic containers gradually replaced the traditional returnable bottles until they virtually disappeared.  The popularity of nostalgia, however, has caused a recent (although limited) revival of returnable bottles in six pack cartons to the grocers' shelves, along with cobalt blue specialty containers and hand-blown bottles. 

     At some point, probably in the 1930s (earlier in some cases), makers of soda bottles began embossing the last one or two digits of the year of manufacture on the bottles, usually on the base.  When I began this analysis, I was skeptical as to the validity of these apparent designations, but, after examining the bases of literally hundreds of post-1940 bottles, I can say that almost all such designations I have seen have fallen easily within the date ranges I have been able to ascertain by other methods.  Embossed year designations, therefore, appear to be generally valid.  However, one- or two-digit dates should be compared with other dating methods whenever possible, as there are occasional discrepancies.  A good example is a Nehi bottle from Liberty Glass Works, Sapulpa, Oklahoma, with the numerals, 61, embossed on the base.  This embossing would seem to indicate 1961, but the bottle has an ACL label that reads, BOTTLED BY ROYAL CROWN BOTTLING CO./EL PASO, TEXAS.  The Nehi-Royal Crown Bottling Company did not change its name to Royal Crown Bottling Company until 1965!  Although the possibility exists that the name was enameled incorrectly, the reliability of the two-digit dating system on Liberty bottles is thereby questioned. 

     These designations, however, seem to become popular about 1942 (although they are found occasionally on bottles from the 1920s and 1930s) on ACL bottles and appear on the vast majority after that date.  Only a few ACL bottles lack year designation on the base (notably those manufactured by Chattanooga Glass Company and Knox Glass Company).  The only late 1930s dates I have seen were on embossed bottles (debossed in the case of Dr. Pepper).  Bottles that have 1930s copyrights on their labels (e.g. Royal Crown, cpr. 1936; Botl-o, cpr. 1939; and Grapette, cpr. 1939) do not bear embossed year designations.  This, along with an extreme scarcity of two-digit dates that begin with "3," suggests that the use of baked enamel labels did not become common until the early 1940s.  It is likely that national brands pioneered the use of ACL labels prior to the early 1940s; smaller local bottlers were probably slower to adopt the newer style of bottles (although El Paso's Empire Products Corporation used polychrome and bichrome bottles as early as 1936!).  Where possible, rather than probable, dates are required, the initial date of 1934 must remain. 

     On the national or wholesale level, bottles can be dated for franchisers who have standardized their containers nationwide.  Coca-Cola, for example, nationalized the famous Georgia green, hobble-skirt bottle beginning in 1916, although it took several years (until ca. 1920) for the transition to become complete.  From that date onward, the dating of Coca-Cola bottles requires a combination of national and local information.  Bottle chronologies for most national franchisers are difficult to obtain, especially in a form usable to an archaeologist.  I have tried to include national chronologies based on empirical or historical data whenever possible. 

     Although Coca-Cola continued to use their famous logo on their bottles with only minor alterations, other franchisers changed logos frequently.  Dr Pepper, for example, went through at least five alterations in their lettering style between 1923 and 1971.  In 1950, the period after the abbreviation, Dr., was dropped, changing the name to Dr Pepper.  Although less frequent than the identity-altering, Dr Pepper, and often minor, logo changes have been a part of most national bottlers. 

     Sometimes patent or copyright dates can offer important information.  A notable example is the paper label on a Mission Orange bottle with a copyright date of 1929.  The label identifies the parent company as the California Crushed Fruit Corporation.  The company became a franchiser, Mission Dry Corporation, in 1933.  The knowledge of the copyright date on the paper label creates a four year span for the dating of amber (almost black glass) Mission Orange bottles. 

     The majority of the bottle information in this work is aimed at dating soft drink bottles on a local level.  Local companies are especially datable when they change names.  Empire Bottling Works, for example, began as a merger between two other companies, Houck & Dieter and Purity Bottling and Manufacturing Company in 1912.  Therefore, all bottles produced by Empire cannot predate 1912.  In 1924, the company was restructured and renamed Empire Products Corporation.  However, because of the national trend in ignoring local bottling information, only the name Empire was used until the change to ACL labeling around 1936.  When Nell Gardner sold out to Richard Price of Grapette in 1956, the new owner labeled the bottles Empire Bottling Company (until the dissolution of the company in 1969). 

     Color change can occasionally determine a temporal difference.  When Nicholson Bottling Works originally went to ACL bottles around 1950, the entire container was labeled in blue enamel.  Around the 1967, Nicholson changed to blue and white ACL lettering.  Many companies used a single color labeling when the ACL process was initiated but changed later to bichromatic or polychromatic labels. 

     One of the most typical methods of dating on the local level is by company longevity.  The majority of the sixty-four El Paso non-alcoholic beverage bottlers were in business less than ten years, with at least eight entrepreneurs only surviving a single year.  This lack of endurance makes their bottles eminently datable.  Bottles from other short timers, however, can easily be dated to the length of their business longevity.  Union Bottling Works, for example, was in business from 1916 to 1935.  Although the company used at least three embossed bottling styles during that time, none are datable beyond that nineteen year span.  Only a single style has been found (so far) for Crown Bottling Works, in operation from 1916 to 1921. 

     Patent dates are usually informative, as are patent numbers, both of which appear occasionally on bottles, usually as an embossment on the base.  Although few researchers make use of the technique, patent numbers can be used to reveal dates.   The use of both date and number can establish an initial date, earlier than which the bottle could not have been manufactured.  The case of the famous Georgia green, hobble-skirt bottle produced by Coca-Cola provides an excellent example.  The original bottle bears the inscription BOTTLE PAT'D NOV 16, 1915 and establishes a date before which the bottle could not have been made.  The date, however, does not tell the complete story.  Few products are put into production immediately after they are patented.  The hobble-skirt Coke bottle was not put into production until the year after its patent and was not accepted universally by Coca-Cola bottlers until about 1920 (Munsey 1972:57-58; Kendall 1978:7).  Patent dates do, however, create and absolute (possible) bottom date for a bottle style. 

     Copyright dates provide another direct dating source.  Like patent dates, copyrights establish a firm date before which a bottle style could not have existed. The earliest Royal Crown Cola bottles contain the message COPYRIGHT 1936 in the left corner of the front enameled label.  Although the existence of Royal Crown was announced in 1935, the bottled product may not have actually been marketed until the following year.  Because this was a new product, the copyright date is likely a very accurate beginning date for the date range of bottle use.  The 1929 copyright on the paper label of a Mission Orange bottle establishes a date for the beverage as a product of the California Crushed Fruit Corporation prior to the establishment of the Mission Dry Corporation in the 1930s. 

     Dates often appear as part of the manufacturer's information on the bottle base, especially after World War II.  Although these appear to be generally accurate, they should be used in conjunction with other corroborative evidence whenever possible.  The earlier hobble-skirt Coca-Cola bottles contain specific dates of manufacture, usually embossed on the heel or lower body of the bottle (Munsey 1972:59). 

     Similar information appears to the right of Owens Illinois Glass Company manufacturer's marks.  Toulouse (1971:403) notes that numbers appear to the left, right, and below the logo.  The left number indicates the factory that made the bottle; the right digit or digits signify the year of manufacture; and the numerals below denote mold information.  In addition, the term "Duraglas" in script is used to designate a new, stronger glass process that was introduced by the company in 1940.

     In discussing the Owens Illinois dating scheme, Giarde (1980:80) states that "the digits were used on returnable bottles so that the company had a means of monitoring the life expectancy of its bottles."  According to Giarde, the company noted that single digits were used during the 1930s, and double digits were adopted during the 1940s.  He then suggests that the company information is not entirely correct (as most of us who catalog Owens Illinois bottles are aware).  He asserts that the single digit was maintained through 1945 with double digits "commencing in 1946" (Giarde 1980:81).  Since the Duraglas mark began in 1940, it adds a new dimension to the dating game and provides a way to distinguish 1930s single digits from 1940s single digits.  Giarde (1980:82) further suggests that few (if any) bottles made in 1929 actually bore the I-in-an-oval-superimposed-on-an-elongated-diamond mark because the merger was so recent.

     Empirical evidence provided by containers of local and national brands of soda suggest that there is a finer distinction that has hitherto gone unnoticed.  Beginning in 1940, Owens Illinois added a dot behind the single digit date.  Therefore, date codes of 0., 1., 2., 3., 4., and 5. indicate the years of 1940-1945.  Single digits without the dot are probably from the 1930s.  A few words of caution should be added, however.  First, the change occurred during 1940.  Therefore, some of the 1940 containers are marked with a dot and some are not.  There are soda bottles with Applied Color Labeling (ACL) embossed with the single digit, 0 (no dot).  The use of the ACL process began on soda bottles in 1934, so these containers must have been made in 1940, not 1930.  Secondly, all of the Owens Illinois plants did not convert to each new system simultaneously.  Double digit soda bottles occur as early as 1944, but the switch seems to have been completed by 1946.  At least one bottle (produced in plant #9) obviously has a "4" (to the left and above the "6") added to the "46" date designation as an afterthought.

     One other major date discrepancy exists on Pepsi-Cola bottles made by Owens Illinois in the mid-1940s.  An example is a bottle with the base embossed 14A49/9  I-in-an-oval-superimposed-on-an-elongated-diamond.  The normal area for the date mark is unembossed.  However, the "49" fits the bottle style as the correct date.  Other examples of the same style include possible date marks (either above or below the logo) of 44, 47, and 53.  In each case, the possible date mark fits into the known bottle style for that time period (see Chapter 7d for the Pepsi bottle dating scheme).  As far as I can tell, this marking style is unique to Pepsi bottles.

     According to Toulouse (1971:395), the number to the left of the logo represents the plant at which the bottle was manufactured.  From looking at El Paso soda bottles, it appears that plants 9 and 6 were the most active in making soda bottles from 1929-1954 (although soda bottles were also made at plants 3, 18, 19, and 23), and plants 5, 7, 9, 15, and 20 made soda bottles after 1954 (from marks on the bases of El Paso soda bottles).  Toulouse states "plants nos. 1 through 20 (except that there were no plants 5 and 19) were active at the time of formation of Owens-Illinois in 1929."  He notes that several plants ceased operation during the Great Depression (mid- to late 1930s), but new plants began operation late in the decade or early in the 1940s.  Empirical data seem to support all of Toulouse's plant information except that of the unused numbers.  Even though there were no plants 5 and 19, those numbers appear (on clear embossing) on El Paso soda bottles.  Two different bottles bearing the I-in-an-oval-superimposed-on-an-elongated-diamond manufacturer's mark bear a "19" in the space to the left of the logo.  Similarly, one container with the later I-in-an-oval mark shows a distinct "5" to the left of the logo.  It is unlikely that Toulouse completely missed two glass factories, so the probable explanation is that those marks had some other meaning to the employees who created them.

Table 1. Owens Illinois Plants Manufacturing Soda Bottles*, Logo #1, 1929-1954**

     Several glass houses included double-digit numerals, such as 46 or 52 embossed at various locations on the base.  These marks may generally be considered dates of manufacture but should still be verified by other dating methods whenever possible.  My empirical research, however, indicates that corroboration is consistent to at least the 95% confidence interval. 

     An interesting twist in dating bottles was used by the Glenshaw Glass Company from 1935 to 1958.  Glenshaw embossed a distinctive G-in-a-square on the base of each bottle to identify themselves as the manufacturer and applied another embossment, a single letter of the alphabet, on the crown finish below the bottle's lip.  Each letter was applied only for a single year beginning with "G" (Glenshaw) in 1935, "H" in 1936, "I" in 1937, etc.  In 1953 ("Y") the numeral "53" was also embossed on the base, a practice which continued until 1958 ("D") when the letter system was discontinued.  Bottles manufactured after 1958 continued to use the two-digit numbering system on the base (Sweeney 1995:75).  An example that used this dating system points out the danger in assuming that spare embossments of two numerals (e.g. 42) indicate dates of manufacture.  A whistle bottle manufactured by Glenshaw Glass Company (author's collection) bears a "Y" on the finish that indicates the year 1953.  The opposite side of the finish, however, is embossed with the numerals, 56, which could easily lead a researcher into selecting 1956 as the year of manufacture.  To further complicate the dating process, the base is embossed with the numerals, 42!  The natural inclination would be to accept the larger embossed basal designation and assume the bottle was made in 1942, an eleven year discrepancy from the actual date of 1953. 

     Two examples should demonstrate the problems with accuracy in starting dates.  Hawkins Dairy (listed as Mrs. L. W. Hawkins prior to 1912) claimed an establishment date of 1902 in their advertisements (in the directories), but were not listed in the El Paso City Directory until 1904.  Nor is the problem restricted to early editions.  The El Paso Herald Post reports the Seven Up Bottling Co. of El Paso as initiating business in the city in 1936, the telephone directory lists the company in the Winter 1937-38 edition, but the city directory fails to mention their existence until 1938.  As illustrated by these examples, however, the city directories are generally reliable to within about two years. 

     A second problem centers around name change.  In some cases, either a name change was not reported to the directory publisher, or a bottler used a different name on his bottles prior to an actual name change.  A case in point is the 1951 and 1952 bottles filled by the Royal Crown Bottling Co. of El Paso and Phoenix.  According to the city directories, the company was named Nehi Royal Crown Bottling Co. from 1941-1965 when the name changed to Royal Crown Bottling Co.  Yet at least two bottles from 1951-52 are marked Royal Crown Bottling Co. more than a dozen years prior to the name change in the directories!  This may be due to an earlier name change of the Phoenix plant, but it seems strange that bottles marked only with the El Paso designation have Phoenix plant's name. 

     Advertisements are only a fair indicator of when a new brand first enters the market and can be inadequate for determining when one passes out of favor.  Coca-Cola ads illustrate the point.  The Coca-Cola parent company introduced Sprite and Fanta in 1960 but advertisements for the two did not appear in the city directories until 1963, two years later.  Mr. Pibb, a Dr Pepper taste-alike, was first marketed in 1972 but did not see directory advertisement until 1977--a five year discrepancy.  Of course, it is possible that the local company did not begin bottling the new drinks when they were first offered by the parent company.  The only ads Nicholson Bottling Works placed in the El Paso Telephone Directories, in 1956 and 1957, did not even remotely reflect either the introduction or the termination of the various Nicholson flavors or King Kola. 

     Until the mid-twentieth century, similar discrepancies applied to listings in telephone directories.  Telephone listings were restricted to those businesses (and individuals) who owned telephones (and chose to list their numbers).  After mid-century, of course, almost any business, regardless of size, was required to have telephone communication to remain solvent.  For twentieth century listings, a combination of both telephone and city directories can improve accuracy.  Prior to the availability of telephone directories (1925 in El Paso), however, the city directory is frequently the only means of demographic information on many firms. 

     Frequently, local dates for manufacturing processes will differ from national ones as two examples from El Paso will demonstrate.  The first example is the crown closure.  Invented by William Painter in 1892, the crown cap was slow to be accepted by bottlers and the public, in part because of the panic of 1893 which prevented many investors from financing a new endeavor.  It has also been speculated that the simplicity of the crown closure "seemed too good to be true (Lief 1965:92).  Riley (1958:102), however, suggests that the slow acceptance was caused by the large number of already established competitors.  Soda bottlers were hesitant to replace machinery that was doing a satisfactory job of sealing their products with something new that would require a revision of their methods along with the purchase of new machinery.  Even when bottlers were willing, containers were not always available.  Glass houses faced the problems of retooling and using up their supply of bottles containing other types of closures before they were willing to begin marketing containers with crown closures.  It was not until 1897 that significant numbers of glass houses were offering crown closures to the trade. 

     In El Paso the magic date appears to be no later than 1900 and possibly as early as 1895.  Early Houck & Dieter bottles featured Hutchinson stoppers as did most R. F. Johnson containers.  When Johnson moved to California in 1895, Henry Pfaff managed his business and changed to crown capped containers.  Pfaff succeeded Johnson in 1898, and the industry changed.  While earlier bottles are scarce, Pfaff bottles are relatively common and, without exception, they are topped with crown finishes.  I have yet to find a single El Paso soft drink bottle that dates later than 1900 which contains other than a crown finish, and it is certain that Pfaff changed the R. F. Johnson bottles prior to 1998. 

     A second example concerns the change in manufacturing technique. Although soda bottles were manufactured by automatic bottling machines as early as 1907 (Riley 1958:106), containers using that technique were not in general use in El Paso until much later. Soda bottles were manufactured using the two-piece mold process (blowing by hand into a mold and then creating the finish using hand tools) from the earliest imported soda bottles until about 1915 (Paul & Parmalee 1973:25).  Bottles from Purity Bottling & Manufacturing Co. (1906-1912) were all made by the two-piece mold process.  The bottle style used by Empire Bottling Works from 1912 to 1922 is found with marks indicating both the two-piece mold process and machine made bottles.  One of the Magnolia Bottling Co. container styles also bridges this gap between glass blowing and machine manufacture with a bottle that was used after 1911 and as late as 1916-1920.  The only known Crown Bottling Works bottle (1916-1921) was made by machine.  Bottles filled by the Mexican Bottling Works (La Mexicana) (1917-1919) used the two-piece mold process.  Texas Bottling Works (1918-1934) used machine-made bottles.  Lone Star Bottling Works bottles (1918-1929) were also machine-made as were all known Union Bottling Works (1916-1935) containers. 

     Mexican Bottling Works may have bought their bottles in Mexico.  Although the company was based in El Paso, none of the owners was listed in El Paso city directories, and they may have lived in Ciudad Juárez.  With that exception, the two-piece mold process was used exclusively by El Paso bottlers until at least 1912.  The switch to bottles made by bottling machines was in place no later than 1916.  The window for change is therefore between 1914 and 1916, supporting Paul and Parmalee's contention that most of the industry had not switched until about 1915 (1973:25). 

     Early bottlers relied on the honesty and understanding of their customers and competitors to ensure the return of these multiple-use bottles.  It soon became apparent that this course was neither wise nor prudent.  In places like pre-railroad El Paso, bottles were scarce and were usually reused until they became broken (Lockhart & Olszewski 1994:47-48).  Such customer reuse meant that containers were often not returned to the bottler.  In an attempt to induce retailers to return containers in the early years of the twentieth century, Houck & Dieter (El Paso's earliest bottler) sent its customers printed postcards that pled: 

 Your account on our ledger shows that you still have in your possession.............of our Mineral Water cases and bottles, shipped [to] you on various dates.  The express company that carried the goods to you, is obligated to carry the empties BACK free of charge to you.  If our mineral water business is to yield us a profit, it is necessary that we re-fill all our bottles and cases as often as possible.  Will you not assist us by causing these cases and bottles to be returned to us without delay? (Postcard in the collection of David Cole, Bangs, Texas) (see Figure 2-2) 

     The solution to the problem of the need for enforced return was the paid deposit.  The idea of a standard deposit on bottles was discussed at the annual meeting of the American Bottlers' Protective Association in 1901, and was endorsed by the Association in 1902.  Paid deposits, however, were not universally brought into effect until the late 1920s Paul & Parmalee 1973:25).  In 1934, the National Recovery Administration required mandatory deposit of 2¢ per small bottle and 5¢ for larger (usually quart size) containers (Woodroof & Phillips 1974:199).  Although the amount of deposit varied at different times and locations, the outlay of money was sufficient to ensure that most people would return the containers.  Because proof of purchase was not required in order to redeem returnable bottles, there was great incentive for children and unemployed adults to collect discarded bottles and insure a fairly steady continuing cycle.  In an attempt to educate the public, early bottlers often embossed their bottles with messages, such as THIS BOTTLE IS NEVER SOLD or RETURN FOR DEPOSIT.  This method was deemed so effective by many bottlers that it continued in use until the 1980s. 
 

Figure 2-2 - 1904 Houck & Dieter Postcard asking for the return of bottles

     The problem with container reuse by other bottlers was not as easy to solve.  From the initial use of the two-piece mold process, begun about 1845, mineral and soda water bottlers often embossed their bottles with their company name and city of origin.  The use of the plate-mold process (explained earlier in the chapter) improved the cost, allowing more of the smaller bottlers to emboss (and thereby insure the return of) their bottles.  The drawback to embossing, especially with the limited area offered by the plate-mold, was a greatly reduced message that was difficult to change.  During the early twentieth century, paper labels were cheap and easy to use, so many bottlers transferred to the new medium.  The initial problem now resurfaced--other bottlers could soak off the labels and reuse anyone's bottle. 

     Two solutions presented themselves and were used by the two largest cola rivals, Pepsi-Cola and Coca-Cola.  The Pepsi solution was to combine both previous processes and place a paper label on a partially embossed bottle.  The solution was so simple and effective for Pepsi that the company continued to use the same standardized style after the baked enamel labeling process became available in 1934.  Replacing paper labels, however, was more of a problem than most bottlers were willing to engage.  Although the washing process removed the old labels without an extra step, applying the new labels required extra labor.  As a result, Coca-Cola and many of the other bottlers returned to embossing, but with new distinctive designs.  This return to embossing created the era of the proprietary bottle, with hobble-skirts, pinched waists, bulging shoulders, stars, bars, orange-peel surfaces, ridges, flutes, diamonds, and all the flourishes that bottle designers could imagine. 

     Proprietary bottles remained the first choice of most bottlers until well after 1934 when the baked enamel labeling process was perfected.  Commonly known in the bottling industry as Applied Color Labeling (ACL), the process fused enamel onto the glass until it weathered almost as well as embossing.  Within a few years almost every bottler, with the notable exception of Coca-Cola who continued to prefer the proprietary bottle (and Dr. Pepper until 1954), had made the transition to ACL (Paul & Parmalee 1973:28; Riley 1958:145, 267).  Many of the national companies immediately chose bichromatic styles for their bottles, usually red and white, the colors chosen by Dr Pepper, Pepsi-Cola (also red, white, and blue), and Seven-Up.  Local bottlers sometimes initially chose the less expensive monochromatic style, then later changed to bichrome bottles.  Nicholson Bottling Works, for example, changed from their 1950s blue lettering to white and blue. Very few companies, such as Pepsi and, for a time, Barq's flavors, actually adopted polychrome labels. 

     Although bottles are often used to establish a date for locations, the application can occasionally be reversed.  A total of 379 complete bottles were removed from the El Paso Coliseum parking lot in the fall of 1990.  The building and parking lot had been constructed atop a city trash dump that was used from about 1910 until shortly before the Coliseum was built in 1942.  Although bottles in the dump could have been (and many were) manufactured prior to 1910, none could have been made after 1942.  The ability to date bottles is therefore often only limited only by the knowledge and ingenuity of the researcher. 

     Because applicable dating involves a multiplicity of techniques, many can be used in combination to refine a date range or confirm an established date.  Validity of dating is best verified by triangulation whenever possible.  Triangulation in this case refers to corroboration of more than one dating technique to substantiate the initially established date or date range.  Three examples of combined techniques will illustrate the point. 

     The first example is one of El Paso's mystery bottles from the El Paso Coliseum collection.  The colorless container is small (contents, 4 ounces) and embossed with the word, Royal, on the front shoulder and, Purple, on the back, leaving space for a paper label below.  Although I have been unable to discover either a company or soft drink brand named Royal Purple (or Purple Royal), the use of the color purple in the name may indicate a grape drink.  The container also provides other indicators helpful for dating.  The lack of color in the glass (and frequent amethyst hue from solarization) suggests a probable post 1890 manufacture, and the crown finish indicates that the bottle was made after about 1895.  The mold marks extending to the container's lip and the presence of an Owens Ring on the base clearly demonstrate that the bottle was manufactured by an Owens Automatic Bottle Machine after 1907, the date when small mouth bottles were first adapted to the process (and probably later than 1915).  Finally, the location of the artifact at the El Paso Coliseum provides an end date, because the bottle could not have been deposited there later than the Coliseum's construction date of 1942.  The probable date range for this Royal Purple bottle becomes 1915-1942, a span of twenty-seven years. 

     A second example comes from the firm of Houck & Dieter.  The bottles appeared in light blue, aqua, solarized purple, and colorless forms with crown finishes.  The containers were vertically embossed in panels around the body with HOUCK & DIETER/COMPANY/EL PASO,/TEXAS and H&D on the base.  The containers exhibited no manufacturer's mark.  The colors suggest an early date for El Paso between 1890 and 1920, but the presence of a crown finish moves the possible beginning date to 1892 with a more probable date between 1895 and 1900.  The bottle was manufactured by the two-piece mold process that began to be phased out about 1907 but was still in use until after 1930 (and was likely terminated in El Paso soda bottles by 1915).  Excluding historical data, the probable date range becomes about 1897 to about 1920.  Houck & Dieter was in business from 1881 until it merged with Purity Bottling and Manufacturing Company in 1912, providing a solid end date.  There was, however, an earlier crown finish bottle used by Houck & Dieter that followed the same embossed pattern of the earlier Hutchinson-type finish bottle.  This style was probably used for more than a year begining about 1898, so the beginning date for the panel bottle was probably around 1900.  The final probable date range becomes about 1897 to 1912. 

     A later bottle from the Grapette Bottling Company will provide the third example.  Grapette began business in El Paso in 1942, and continued until the company's dissolution in 1969.  This colorless, crown finished Grapette bottle has a bichrome ACL label.  The enameled label, initiated in 1934, negates the need to seek dating information from color (post-1880), finish (post-1892), or manufacturing technique (post-1907).  A faint manufacturer's mark appears on the base, but it is so indistinct as to render it illegible.  The bottle back, however, bears a copyright date of 1952, providing the best beginning date for this bottle type.  A legible numeral, 53, to the left of the manufacturer's mark provides corroboration for the 1952 beginning date.  With no other indicator to establish a terminal date for this bottle type, the final year of business for Grapette in El Paso, 1969, must be assigned. 

     A final situation, also unique to returnable bottles, is frequently encountered.  The United States is, and traditionally has been, a mobile society.  Because of our constant migration, we leave a trail of introduced trash wherever we go that includes returnable soft drink bottles from other locations.  Coca-Cola bottles embossed with EL PASO, TEXAS, are found in Maine.  Barq's bottles from Tacoma, Washington, are located in Baton Rouge, Louisiana.  Bottles from the American South are common in El Paso.  Whenever these foreign bottles appear in an area, local bottlers refill them.  They receive no deferential treatment; in fact, they are not even inspected to determine origin (although they are examined for cleanliness prior to refilling).  Thus, a Dr. Wells bottle transported by a tourist to El Paso from Blacksburg, Virginia, in 1945 would have been returned for deposit in a local store and finally redeemed by Barq's Bottling Company (the local distributor of Dr. Wells) of El Paso.  Rather than attempting to return the bottle to its place of origin, the line workers at Barq's would have refilled the bottle for local distribution even if the bottle style differed from the configuration used locally.  Locally marked bottles, however, can be expected to greatly outnumber those of foreign origin. 

     Deposition by tourists and travelers, along with occasional importation by retailers for local distribution, also accounts for the intrusion of foreign bottle types that were not distributed within the region.  Soft drink bottles from as far away as Albany, New York, and Belfast, Ireland, were introduced into El Paso either prior to or concurrent with the establishment of Houck & Dieter, the city's first soft drink and mineral water bottler in 1881.  Bottle collectors excavated unmarked cork (blob top) and Hutchinson stoppered bottles at one of El Paso's early dump sites (Chamizal).  These late nineteenth century bottles were not bottled locally and were therefore imported by one of the above means. 

Table 3.  Precision of Dating Methods

     Despite the required deposit and their inherent strength, returnable bottles were occasionally either accidentally or intentionally broken (e.g. as a target in shooting practice).  Fragments appear in the archaeological record as a result of primary deposition (i.e. leaving the fragments where they fell) or secondary deposition (i.e. cleaning up the debris and redepositing them in the normal manner of trash disposal) depending on where the breakage occurred (Schiffer 1987).  Unlike breakage caused by consumers which was likely discarded on an individual basis, breakage incurred by retailers (returned to the bottlers) and the bottlers' own breakage was generally curated and later deposited in quantity.  Depending on temporal and physical location, deposition could occur in an established landfill, casual dumpsite, or private site (usually a pit) within the bounds of the bottler's property.  A casual dumpsite in Socorro, Texas, just southeast of El Paso, for example, contained a pile of glass shards at least two meters in diameter.  The surface of the pile disclosed no complete containers, although all fragments came from bottles used by the Lone Star Bottling Works of El Paso.  Onsite dumping was practiced frequently throughout both the city and county of El Paso by homeowners and renters as well as commercial businesses.  Usually, trash (including broken bottle glass) was stored temporarily then deposited in an excavated area generally located behind the dwelling or business.  Richard Price told of finding a large pit of broken (and some complete) beer bottles when he supervised the reconstruction of the old El Paso Brewery in the 1970s (Price interview, 1996). 
 <place photo of Socorro Dump about here> 

     Onsite deposits also contain foreign brands (i.e. brands not belonging to the individual bottler).  These usually arrived mixed into a case of legitimate bottles that were not checked carefully enough by the route salesmen delivering product and collecting empties.  The bottler was left with a choice of informing his competitors that he had received their bottles, returning them to their correct owners, or disposing them along with his own broken or worn out containers.  Although the only local example I have discovered involved beer (El Paso Brewery) instead of soft drinks, the discard of foreign bottles was common, especially around the turn of the century (Price interview, 1996). 

     Occasionally, large depositions of complete bottles are discovered.  When bottle collectors excavated the Chamizal dump in 1965-1968 just before the area was traded to Mexico, they found large quantities of complete bottles from the Purity Bottling and Manufacturing Company (1906-1912).  According to one informant, the bottles were so ubiquitous that many of the collectors began leaving them alongside their excavation pits rather than keeping them (Garrett interview, 1996). Two other early bottles are also common in El Paso.  Containers from Henry Pfaff are frequently found (also excavated in the Chamizal digs) despite his equally short duration as a bottler (December 1898-1907) although bottles of his predecessor, R. F. Johnson (1983-1998) are extremely rare.  Similarly, the third and final style of bottle used by Houck & Dieter (1881-1912) from about 1900 to 1912 are more common than the similar style containers used by their descendant, Empire Bottling Works from 1912 to about 1935 despite the latter being used for a much greater temporal period and more recently. 

     The greater quantity and availability of these three bottle types than all other early ones may be explained in part by the popularity of the brands.  But prominence in the community cannot explain why both the final-style Houck & Dieter containers and the Purity Bottling and Manufacturing Company bottles are more common than the later Empire bottles that were used for about twenty-three years or why Houck & Dieter's third bottle, used for about twelve years is common and the first bottle (used for about nineteen years) is rare (the second bottle was a transitional container, used only briefly).  Additionally, Houck & Dieter had a virtual monopoly during the nineteenth century but sustained added competition from 1900 to 1912.  Why did only these three brands survive well. 

     The answer appears to lie in the concept of unplanned or unanticipated transitions.  Henry Pfaff had been an employee of R. F. Johnson in the mid- to late 1890s and actually managed the business after Johnson returned to California in 1895.  He was in an excellent position to allow almost all of the older R. F. Johnson and El Paso Bottling Works bottles (another Johnson label) to gradually run out while introducing his own brand in 1898.  Documentation pertaining to the sale of Pfaff's firm to Southwestern Liquor Company (1908-1914) is not as complete, although Pfaff employed A. L. Longneckard, the initial president of Southwestern, for two years prior the sale.  The transition was undoubtedly more abrupt than that between Johnson and Pfaff, and Southwestern immediately terminated the Pfaff brand.  The result was a large number of complete, embossed  Pfaff bottles that required disposal. 

     Intracompany transition was also carefully planned.  When Houck & Dieter phased out their Hutchinson-stoppered bottles for their transitional container and shortly-thereafter for their final style, they likely gradually reduced the last of the older style bottles rather than disposing of them.  When Houck & Dieter joined with Purity to form Empire Bottling Works, however, both parent companies, each of whom had continued to produce soft drinks prior to the merger, had bottles with now-defunct brand names that required disposal.  The Empire Bottling Works containers that followed, however, were again transitioned gradually out of production.  The hypothesis, then, can be stated:  Unexpected transitions such as the abrupt termination of a particular brand of soft drink or company name can be expected to result in the deposition of greater quantities of intact soda bottles than would normally be expected from more gradual transformations.

     Adams and Gaw (1977:218-222) suggest a more complicated operation that includes container manufacture, package, wholesale, retail, and purchase, each with a possible accompanying storage lag.   The process continues with use (including possible reuse), discard, and, finally, entry into the archaeological record.  They conclude that time lags are of unpredictable duration.  Although Adams and Gaw are concerned primarily with ceramic artifacts, they generalize their findings to artifacts in general.  Newman had earlier suggested six causes of deposition lag for bottles: 

 1. delays in particular bottle manufacturers changing over to the new technique; 
 2. the use of existing stocks of bottles made by an earlier technique after a new technique was developed; 
 3. storage time at the manufacturer's factory prior to shipment to the bottler; 
 4. delays in filling the bottles with the contents to be retailed; 
 5. transportation time to the site; and 
 6. the possibility of re-use of the bottle prior to final discard (Newman 1970:70). 

Newman concluded that "ten years seems adequate to allow for the time delays in American bottles reaching most archaeological sites in the United States (Newman 1970:70). 

     The length of storage is dependent on the contents of the container.  Wine and whiskey, for example, are intentionally aged by the intermediary dealer (the winery or distillery) either before or after bottling.  Shoe polish, hair dye, and similar products are not time-sensitive and may also be stored for long periods of time by their manufacturer or the retailer--as well as in the home by the purchaser.  Items with a short manufacturer-deposition lag, such as soft drinks, beer, or especially dairy products, would have a very brief expected storage life (Hill 1982:294-298). 

     Occasionally, however, the purchaser (end-user) may reuse the container after the original product has been exhausted.  The bottle may be reused for the same purpose:  historically whiskey flasks were regularly refilled at the local tavern or bar, and home-brew aficionados still reuse bottles that once contained nationally retailed beer.  The objective for reuse may also be entirely different from that of the container's original purpose.  More than one generation during the first half of the twentieth century grew up watching their mothers dampen clothes prior to ironing with a sprinkle attachment affixed to the end of a Coca-Cola or Pepsi-Cola bottle.  During the Prohibition era, so many soda bottles were reused for bottling beer, that it caused a shortage in the soft drink industry (Busch 1991:117).  Virtually any liquid found in a normal household has been stored temporarily or permanently in the nearest convenient glass container, and wine bottles have been notorious for their use as candleholders.  Finally, a container may be curated by the end-user.  Bottles are often saved for aesthetic or sentimental reasons.  Indeed, many containers, notably the Avon collectable selection of colognes and perfumes, are intentionally packaged with the aim of container retention by the consumer.  Bottles may also be curated as a memoir of a special occasion, such as an anniversary, first date, or birthday. 

     An example of a reused bottle is a Pepsi-Cola bottle with a sprinkler top.  The bottle itself was one of the first standardized Pepsi-Cola containers used between 1939 and about 1945 (see bottle descriptions under Woodlawn Bottling Company).  These bottles contained paper labels to convey their advertising message, and remnants of the old label still clings to this example.  The base of this bottle was embossed with the manufacturer's mark of the Oil City Glass Bottle Company used between 1930 and 1952.  The mark is followed by 14 A 47 4/DES. PAT. 120,277 (Robert Bejarano collection).  The numerals, 47, likely indicate the year, 1947, despite being two years after the change of technique to baked enamel labeling (see Woodlawn section for a detailed chronology of Pepsi-Cola bottles).  The stopper, itself, is made from rolled tin with a cork sealant to prevent unwanted leakage.  The head is repeatedly perforated to allow a controlled flow of water droplets.  Prior to the common use of the steam iron, clothing was sprinkled using these containers to wet out wrinkles that were than ironed dry.  A second example is a Georgia Green six-and-one-half-ounce Coca-Cola bottle with sprinkler stopper (Lawrence Angus collection). 
 

Figure 2-3 - Coke Bottle with Stopper	Figure 2-4 -  Pepsi Bottle with Stopper

 In over 90% of the listing, age was not given, suggesting the bottles were fairly new.  One group of bottles which was 3 1/2 years old was listed as being of no value.  Probably less than 1/10 of 1% of the bottles were over five years old (Miller & Pacey 1985:40). 

The probability of a short storage period is justified by several factors:  1) bottlers were fickle, changing suppliers according to market fluctuation; 2) nationally franchised brands changed bottle styles relatively frequently, often in less than ten years; 3) small bottlers were noted for the short duration of their businesses--several El Paso bottlers remained in business for only a year; 4) even relatively stable local bottlers with national franchises were prone to change company names, necessitating a new bottle run; 5) bottlers usually bought a sufficient supply at one time to justify a run of bottles, even on the high-volume Owens Automatic Bottle Machines (The Owens machine required a production run that consisted of a large number of containers to be economically feasible. For a more detailed explanation, see Miller & Sullivan 1984:86).  Any one of these factors is probably not convincing by itself, but, in combination, they suggest a short, cautious storage time for returnable soda bottles by the manufacturer. 

     Empirical data examined by the author suggest a storage period by the manufacturer of not more than one year prior to sale to the bottler.  After the use of the two-digit dating system (embossed on the bottoms of returnable soda bottles) became common practice, a bottle occasionally contains a date that is one year earlier than a bottler's opening date in business.  However, I have never seen a bottle that is dated more than one year prior to the opening year (even one year is very rare).  The above data suggest that glass containers manufactured for returnable soft drinks were generally not produced earlier than one year prior to sale to the bottler and then only in the case where the bottles were of a standard design and labeled with the ACL process (post 1934). 

     The use-life of a one way bottle is quite complex (see above), but that complexity is multiplied by the intentional cyclical reuse of the bottle.  Because a returnable bottle makes multiple trips between the bottler and the imbiber of the contents, the average bottle makes a relatively predictable number of returns.  Various El Paso bottlers have estimated (or used home office estimates) the number of trips between 12 and 52 with the majority somewhere closer to the middle of that range.  In a personal communication, Mildred G. Walker, curator of the Dr Pepper Museum, Waco, Texas, said bottlers informed her that returnable bottles survived about eleven washings because of the caustic solution (NaOH) involved.  The bottle design became eroded by that time along with the lip.  An abraded lip would not seal properly so the bottle was discarded.  Raised lettering on embossed bottles was particularly susceptible to wear.  Busch (1991:122), suggests 24 round trips. An average-selling brand could probably expect to revolve its entire stock (in an average grocery store) at least once a month (less often for more popular brands, longer for less preferred labels) with greater rotation in hot weather, poorer in cold.  In other words, an average bottle made a single round-trip in about 30 days.  The simple use-life of a bottle, therefore, was about a year or less (52 trips multiplied by 30 days = 1,560 days or 51 months or 4 years, 3 months [using the maximum possible number of expected trips]).  By the end of that time, a bottle would be stolen (all returnable bottles belonged to the bottler, not the purchaser, who only bought the product contained in the bottle), broken (by either the bottler or the user), or became so worn that the bottler was ashamed to reuse it and therefore discarded it. [Note: When I first calculated the use-life of returnable bottles, I erroneously assumed a turn-around time of about a week, producing a much shorter expected life. See Lockhart 1999:160] 

     Although rarely addressed, the use-life or number of round trips varies in direct proportion to container size. Many of the older six- or seven-ounce returnable bottles can usually be dropped on a concrete surface from a height of three feet or more without any visible damage (although the finish often breaks off on the third or fourth bounce).  According to Thomas Lucky, lifelong employee of Magnolia Coca-Cola Bottling Company in El Paso, six-and-one-half-ounce bottles made 20-25 trips per bottle; ten-ounce, maybe 14-15; and fewer for sixteen-ounce containers.  Quart bottles, of course, showed an even smaller return.  Woodroof & Phillips (1974:202) noticed the change (although not the reason) when they stated that "the glass manufacturers were noting a rapid decline in the trippage of returnable bottles--from an average of 21 trips per bottle in 1959 to 14 trips in 1969."  Container size further compounds the problem of precise assessment of use-life. 

     The next question is:  how often does a bottler buy bottles?  The answer in a word--frequently (at least for medium to high-volume bottlers).  According to Joe W. Yowell, former owner of Barq's Dr Pepper Bottling Company in El Paso, his business bought six to ten loads of Dr Pepper bottles per year.  Even his worst selling products demanded new bottles at least annually.  Other bottlers concurred.  Richard C. Price of Empire Bottling Company bought bottles at least annually, usually several times per year.  Smaller bottlers or less popular brands may have replaced  bottles less often.  In examining bottle types used by Crystal Beverage Company of Alamogordo, New Mexico, for example, I have only found two dates on the bases: 1943 and 1947 (cf. Lockhart 1998:34).  This suggests that bottles were bought no more often than every four years.  The Crystal plant was noted for large stacks of shells (wooden cases) loaded with empty bottles (Callaway interview, January 23, 1998).  Even larger bottlers sometimes stocked up to take benefit from advantageous situations. Robert Winstead of the Meridian, Mississippi, Royal Crown Cola Company took advantage of free use of railroad boxcars to store a special order of bottles in 1947.  That meant that some of the 1947 run of bottles was not used until 1949 (Elling 1998:[4]). 

     Under normal use then, a bottle from a popular brand could be expected to become part of the archaeological record in about five years (no more than one year warehoused in the bottling plant plus about four plus years of use-life).  Although not as positive, three to four years could be added to a less popular brand.  However, the statement of a bottle detective in New York, hired to investigate bottle theft, suggests an even shorter, two-year or less timespan.  He noted in 1906 that very few bottles still remained in use that were dated 1903 or 1904 (Busch 1991:118). 

     The scant available empirical data support the longer period.  Although the survival of entire cases of returnable bottles is rare, a few are located in private collections.  I bought a case of Bubble Up, a less popular drink, at a local antique store.  This case of twenty-four bottles contains bottles ranging in age from 1939 to 1946 (7 years) with 75% of the dates clustering between 1942 and 1944 (3 years).  Michael M. Elling (1998:[3-4]) provided two cases, one from Coca-Cola, the other, Mellow Yellow.  The Coca-Cola case contains bottles that range from 1987 to 1995 (8 years) with 79.2% of the dates clustering between 1991 and 1995 (4 years).  Bottles in the Mellow Yellow case range from 1975 to 1986 (11 years) with 83.3% clustering between 1975 and 1979 (4 years).  The mean total range is 8.6 years with a mean cluster range of 3.7 years.  Although the sample is very small, the data  suggest about an eight-and-a-half-year possible deposition lag for these bottles, with a probable deposition lag of about four years. 

     The general five-year lifespan suggests that an excavated bottle was likely to have been deposited in the archaeological record within about five years after its maximum datable period (expanded to eight or nine years for less popular brands).  The well-known Christmas Coca-Cola bottle provides a good example to work with.  The Christmas Coke bottles was patented December 25, 1923 but saw no actual use until the following year (1924).  Munsey suggests, based on information from the Coca-Cola home office, that such bottles were in use from 1924 to 1937; Kendall asserts that an overlap should be assumed because it takes time for the bottles to get into circulation after the actual patent process is complete, and the dates actually range from 1926 to 1938 (Kendall 1978:7; Munsey 1972:62-63).  Combining Kendall's system of overlap with Munsey's home office data (to give a maximum possible range), an effective date range for any given excavated Christmas Coke bottle is 1924-1943, allowing five extra years to compensate for the use-life until most of the final bottles have become part of the archaeological record. 

     It is certain that an unknown percentage of bottles are curated for reasons varying from reuse to sentimental attachment or decoration.  Because of the required deposit, however, returnable bottles are less susceptible to curation than nonreturnables.  Nonetheless, the high volume of clean and sparkling returnable bottles gracing collections throughout the United States clearly reveals that people have frequently chosen to retain returnable bottles rather than restore them to their legal owners, the bottling companies.  Perhaps the most common form of curation for returnable bottles is accidental storage.  Numerous cases of bottles that were discovered in garages or attics as much as thirty years after the purchase of their contents have been reported to Mildred Walker, curator of the Dr Pepper Museum in Waco, Texas (personal communication).  Unfortunately, there is no available method to discern or even estimate the number of bottles curated by individuals.   It is probably unusual, however, that curated bottles return to the archaeological record.  Such bottles are increasingly likely to become instead part of either museum or private collections.  The popularity of bottle collecting is evidenced in numerous national and regional collectors clubs and several publications devoted to glass containers (a few of the most popular collector's magazines include Antique Bottle & Glass Collector, Bottles and Extras [Federation of Historic Bottle Collectors], and Soda Net [Painted Soda Bottle Collectors Association]). 

     An additional complication arises, particularly in smaller companies or companies bottling less popular beverages.  In some cases, bottlers stored their containers for weeks or even months before bottling a product with a slower turn-around rate.  Often secondary brands were only bottled once a week or less, and that could frequently be delayed during the off-season to two weeks or more.  In these cases, numerical life expectancy (i.e. number of trips) would remain the same, but temporal life expectancy (number of weeks or months) would increase (personal communication, Mildred G. Walker). 

     However, we can assess the logical probability of a bottle's normal deposition into the archaeological record.  Returning to the Christmas Coke bottle as an example, two sales records for El Paso are available from the National Archives.  In 1931, Magnolia Coca-Cola Bottling Company sold 95,473 cases of 6½-ounce bottles.  An unknown number of these were Hope's flavors that came in a distinctive "square" bottle.  Square bottles embossed with El Paso, Texas, are at least as difficult to find in collections and antique stores as the Christmas Coke bottles.  I therefore treat the two as a single population. Two years later, due to the Depression, sales had fallen to 65,216 cases.  Both figures fall within the 1926-1938 period of use for the Christmas Coke bottle.  Figuring an average annual sale of 90,000 cases prior to 1931 and an average of 65,000 after that date, total sales of Christmas Coke bottles during the thirteen-year period of their use was 995,000 cases annually or a total of 23,880,000 bottles sold.  Dividing that figure by the maximum estimate for possible bottle reuse (52 trips) provides a figure of 459,230 bottles actually refilled during the use period for Christmas Coke bottles.  If only 5% of those bottles were curated, then 22,961 Christmas Coke bottles bearing the El Paso, Texas location embossment on their bases would have to remain in existence.  Using a more realistic estimate of 26 round trips (a more central figure), 918,461 bottles were used.  A 5% curation rate would then equal 45,923 remaining bottles!  Either figure is excessively high for the actual number of Christmas Coke bottles in circulation based on the availability in the collectors' market.  Thus, if less than 5% of the Christmas Coke bottles were curated, then an archaeologist can maintain a greater than 95% confidence that any given bottle of this type located during an excavation has become a part of the archaeological record by a normal channel (i.e. broken or discarded) and actually dates to the 1924-1940 time period. 

     If we project these findings onto other types of returnable soft drink bottles, an archaeologist may, within a 95% confidence interval, assume that a returnable soft drink bottle found in an excavation is datable to its known use period plus five years for the final production run to exhaust its cycle of round trips.  In the case of post-1942 bottles (and some earlier ones), where two-digit manufacturing dates are embossed on the bottle base, a five-year deposition period may safely be assumed within the 95% confidence interval.  To be on the safe side, an additional two-three years may be added to less popular soda brands to account for additional delays (see above).  Soft drink bottles, therefore, provide a very precise dating format for late nineteenth and twentieth century sites that are still within the federally-prescribed historic period.  Broken (but datable) bottles provide an even greater confidence interval since curated bottles would be less likely to be broken than ones in normal use. 

     A second way of viewing attrition is provided by Woodroof and Phillips (1974:431)   who suggested that about 1950 "bottles were returned and refilled, often with a loss rate of only one bottle per case-trip."  Assuming a twenty-four bottle case (the norm in 1950), it would take 24 round trips before the entire case was lost to either theft or breakage.  This view supports the more realistic twenty-six round trips suggested above. 

     According to Richard Price, one-time owner of the Grapette Bottling Company and Empire Bottling Company in El Paso, there was another outlet for bottles in border communities that was not problematical for non-border areas.  Price suggested that a large number of bottles were either casually or intentionally exported into Mexico and refilled there.  That problem would primarily apply to brands produced by bottlers on both sides of the Rio Grande, such as Coca-Cola, Pepsi-Cola, or Barq's flavors, but casual deposition in Mexico (e.g. discard by tourists) could also pertain to other beverages bottled only on the United States side of the border.  Bottles could be reused or casually discarded rather than putting forth extra effort to return them to their country of origin.