Three Rivers Glass — Maker's Marks

The Three Rivers Glass Company marked its machine-made bottles and jars in several forms, always anchored by a five-pointed star — “the only constant in identifying a piece of Three Rivers glass” [@ev-0038, p. 33; @ev-0039, p. 33]. Five variants are documented [ev-0042]. This article is the reference set for the mission’s first question, and its dating rules are what a mark can and cannot establish on its own.

Dating rules

Before 1925, no mark. The company made its glass by hand in its first years, and marks have been observed only on machine-made ware: “Apparently, products manufactured prior to 1925 were unmarked” [ev-0042]. An unmarked Three Rivers piece is therefore not evidence against Three Rivers manufacture, and this is the single most common way the marks mislead.

From 1925 to 1937, the star marks. First use coincides with the firm adopting machines in early 1925; the end point is 1937, when Ball Brothers finished filling the orders outstanding when they bought the company [ev-0042]. Except for the probably-early all-letters form, current evidence cannot date any variant more narrowly than this full span: the variations may reflect dates, sizes, or glass styles, or may “only have been changed at the whims of the mold makers” [ev-0042].

From at least 1928, two-digit date codes on the heels of milk and soda bottles, and possibly on all bottles; the documented example is a peppersauce bottle carrying a 1929 code [ev-0042].

Exactly 1937, the dual-marked Ball bottles. A bottle carrying the 3 RIVERS {star} logo on the heel and the cursive “Ball” mark on the base can only have been made during 1937, while Ball Brothers ran the plant to fill existing Three Rivers contracts. An El Paso soda so marked is documented [ev-0042], and Smith’s price list independently records the combination on an “Ideal Bottling Co.” soda, noted there as “very unusual” [@ev-0038, #37; @ev-0039, #37]. This is the one case where two marks on one bottle date it to a single year.

Why the marks are faint

Smith advised collectors that “some logos will be somewhat worn away” and recommended holding a suspected piece up to strong light [@ev-0038, p. 33; @ev-0039, p. 33]. The Bottle Research Group gives a different mechanism: in their experience most marks were weakly struck rather than worn, because the logos sit on heels and bases where mold lubricant accumulates, and a mold not cleaned regularly could build up enough residue to render a mark indistinct or obliterate it entirely [ev-0042].

The two accounts disagree about cause and agree about practice. Either way a faint mark is expected and strong light is the method; but the distinction matters for attribution, because a weakly struck mark was faint when the bottle left the plant and tells nothing about how the bottle was used since.

The five variants

THREE RIVERS {star} — possibly 1925–1927, earliest

The full company name with a star, on heels or bases. Only a couple of examples are documented: a milk-bottle base with “THREE RIVERS” arched around a comparatively large star, roughly five times the size of the stars in later logos and partly obliterated by the ejection scar, and a soda bottle with “THREE” arched over the star and “RIVERS” in an inverted arch below. This is judged “probably the earliest mark used by the glass house, possibly 1925–1927 — although the dates are an educated guess” [ev-0042].

Toulouse recorded the mark without the star and dated it “1927 to 1935, rarely used”; Giarde noted the mark on milk bottles. Neither work is held by this expert, and both are known here only as the Bottle Research Group reports them [ev-0042].

3 RIVERS {star} — 1925–1937, most common

The numeral form, the most common configuration observed, occasionally accompanied by a letter, a number, or both [ev-0042]. Smith’s own text names this variant the most common: “The most common logo variation was ‘3 RIVERS,’ in all capitals, followed immediately by a five-pointed star” [@ev-0038, p. 33; @ev-0039, p. 33].

On Mason jars the logo runs in an arch with a design number; on bottles it was often embossed horizontally. The arched jar form is reported from Toulouse (“‘3 RIVERS’ in a half circle, with a number, 107, indicating a design specification”), from Roller, and from Creswick, who illustrated a Mason embossed “3 RIVERS {star} (arch) / 511 / 4” — all three known here only through the Bottle Research Group’s account of them [ev-0042]. The mark is documented on the heels of at least two El Paso sodas, one of them the 1937 dual-marked Ball bottle above, and on the heel roll of pint milk bottles made for the Mistletoe Creamery of El Paso [ev-0042].

Because this is the most common variant, the full 1925–1937 range fits it best; other variants may have run shorter [ev-0042].

3R {star} — 1925–1937, bases only

The monogram form. Smith called it common, found only on bases, and frequently followed by a single-digit number, often a “7” [@ev-0038, p. 33; @ev-0039, p. 33]. The Bottle Research Group, whose sample was mostly soda and milk bottles, found a single example and suggests it may have been used more on smaller bottles [ev-0042].

Smith’s price list entry #52, a 15-oz. “No. 1 in Quality and Size” root beer bottle, records a piece marked twice — “3R{star}” on the bottom and “3 Rivers {star}” on the side [@ev-0038, #52; @ev-0039, #52]. That is direct evidence the two forms were in concurrent use on the same ware, and it is the reason a base monogram and a side name mark do not indicate different dates.

{star} 3 RIVERS {star} — 1925–1937, El Paso Union Bottling Works

Stars at both ends of the numeral form, documented on two different bottles used by the Union Bottling Works in El Paso, one with the mark on the heel and the other on the base [ev-0042]. Smith does not list this variant [@ev-0038, p. 33; @ev-0039, p. 33].

THREE {star} RIVERS — 1925–1937, least common

The star between the two words. Smith reported it as rarely seen but occurring on either heels or bases [@ev-0038, p. 33; @ev-0039, p. 33]; the Bottle Research Group has never found an example [ev-0042]. This variant rests on Smith alone, and no independent observation of it is held here.

Mason jars, packers’ jars, and what a mark cannot settle

Three of the works named above report Three Rivers Mason jars: Toulouse’s “‘3 RIVERS’ in a half circle, with a number, 107, indicating a design specification,” Roller’s arched form, and Creswick’s illustrated Mason embossed “3 RIVERS {star} (arch) / 511 / 4” — all reaching this article through the Bottle Research Group [ev-0042]. The plant’s own chief chemist, asked under oath whether Three Rivers manufactured any fruit jars, answered: “No, not so-called. They were fruit-packers ware, one trip container; they weren’t Mason type jars” [@ev-0059, part 04, fol. 3430].

That looks like a flat contradiction and is not one. The same record shows how a single jar satisfies both descriptions. In January 1935 Hartford wrote to Frank C. Ball, who had reported Mason fruit jars in the trade that he believed Three Rivers had made, and recounted what the receiver had said when summoned to Hartford: he had never manufactured fruit jars for the domestic trade, but “did admit that on orders received from one or two packers he had supplied a straight line jar with a screw top thread,” the packers having ordered them for packing, and if a buyer then “had gone out and bought caps, and had then sold the jars and the caps as a domestic fruit jar, such action was beyond his control” [@ev-0059, Gov. Ex. 203, part 16, fol. 59030]. Coleman conceded the same end use in the same breath as his denial: asked whether the company sold its packers’ ware to be used as fruit jars, he said “Well, I think Mr. Ball said they did” [@ev-0059, part 04, fol. 3430].

So the disagreement is about trade and licence, not about glass. A jar with a screw-thread finish, marked 3 RIVERS {star}, is correctly catalogued by a collector as a Mason-finish jar and was correctly called one-trip packers’ ware by the firm that blew it. Neither party is mistaken about the object. For identification the rule is narrow and worth stating plainly: a mark tells you the maker and nothing about the ware class, and a Mason-finish Three Rivers jar is not evidence that this expert’s sworn source is wrong, nor is Coleman’s testimony evidence that a Mason-finish Three Rivers jar is misattributed.

And the licence line is in the record. Hartford kept a tabulated register of its licences, printed in the trial record as a defendant’s exhibit, giving each licensee’s permitted ware field. Three Rivers’ entry — left margin shaved by the scan — runs: “[Three R]ivers Glass Co. - Three Rivers, Tex. / [bever]ages, [pa]ckers, proprietary and [m]iscellaneous ([she]llacs, shoe polishes and bluing), pressure, beers, still waters, liquors, fruit juices” [@ev-0059, part 18, line 67552]. Packers’ ware is granted; domestic fruit jars are not, and that class was available to the compiler, since the entry immediately below — Illinois Pacific Coast — ends “domestic fruit jars.” (“Fruit juices” in the Three Rivers line is a beverage class, not a jar, and is easy to misread as the thing it is not.) So Coleman’s “they weren’t Mason type jars” was not one witness’s way of putting it. It was his employer’s licence, stated from the stand, which is why he could deny the ware class and concede the end use in consecutive answers without contradicting himself.

That sharpens the rule above rather than replacing it. A mark still tells you the maker and nothing about the ware class — but where the company’s own words about ware class survive, they are the words of a licensee describing what it was permitted to make, and should be read as such rather than as a description of the glass.

One dating rule this suggests and does not establish. In the same 1935 letter Hartford told Ball that if the court forced it to reinstate a licence with a reorganized Three Rivers, it would “make every attempt to write into that license agreement with Three Rivers Glass Co. that they will not make jars with the Mason finish and seal” [@ev-0059, Gov. Ex. 203, part 16, fol. 59030]. If such a clause was ever written, a Mason-finish Three Rivers jar would date before 1935. Nothing held here says whether it was, and this is recorded as a testable proposition rather than offered as a dating rule.

The register named above is the document class that would record such a clause, and it has now been read for one. Besides setting out ware fields it keeps a dated ledger of changes to them, licensee by licensee: “3/30/36 - Maywood - limits on jam jars increased,” “6/15/36 - Maryland - limits on pressed j[a]rs removed” [@ev-0059, part 18, line 68242], and the same form seven years earlier — “9/6/29 - Olean Glass - grape juices, cherries, olives,” “10/10/29 - Glenshaw

• citrates of magnesia” [@ev-0059, part 18, line 67552]. A clause forbidding a Mason finish is a limit on a ware field, and this is the book where limits on ware fields were written down. In the parts of it this expert holds there is no amendment to the Three Rivers field of any kind; the company’s single line in the ledger is “11/27/36 - Three Rivers - acquired by Ball.”

That is a search result, not an answer, and the reason is worth stating because it bounds every future attempt on this question from this source. The extract holds only passages naming “Three Rivers” or “Searcy,” with twenty-five lines on either side. The exhibit is a table about dozens of licensees, so it survives here only where this company happens to be named: three short windows, the first holding little more than its ruined title line and the other two holding ledger, each separated from the next by six or seven hundred lines the extract never captured. No amount of rereading recovers what lies between them. That is an acquisition, not a reading, and it is entered as one below.

After the Ball takeover: the “HA” variant

Creswick listed a jar-base variant reading “3 RIVERS {star} HA” and attributed it to the Hazel-Atlas Glass Co., the “HA” almost certainly being the Hazel-Atlas monogram — reported here through the Bottle Research Group, which draws two inferences from it: that the jar was intended as a product or packers’ jar rather than a fruit jar, and that the mold was transferred from the Three Rivers plant to Hazel-Atlas, probably after Ball Brothers took over. Because packers’ jars have never been comprehensively studied, all or most of the Three Rivers product molds may have gone the same way — though the “HA” could instead be some form of mold code [ev-0042].

The transfer inference has documented precedent in Hazel-Atlas’s own practice: early “-Atlas-” Mason jars whose Atlas embossing is “often lighter than the rest,” suggesting it “was cut into acquired molds,” and Whitney Mason jars with ghosted ATLAS/WHITNEY embossing implying Atlas ran re-cut Whitney molds [ev-0043]. That is a second, independent body of evidence that the firm did exactly the thing a “3 RIVERS {star} HA” jar would represent — which raises the inference above speculation without establishing it.

Hazel-Atlas had been watching this plant’s jars. In March 1933 an interoffice letter out of Dallas, addressed to Hazel-Atlas Glass Co. at Wheeling, West Virginia under the subject line “Three Rivers,” reported that the company was “now offering a regular packers jars in Pint and Quart sizes either Round or Square equipped with two piece screw cap to jobbers in the Sherman district,” gave the prices offered on quarts and pints, and warned that “if they do this generally in Texas it will certainly affect our fruit jar business this coming year” [@ev-0059, Gov. Ex. 945, part 16, fol. 60283]. That is not evidence that a mould moved. It answers the unspoken objection to the transfer inference — that a large eastern manufacturer would have had no particular interest in a small Texas plant’s jar moulds — because Hazel-Atlas was tracking this exact line of ware by size, shape, closure and price four years before Ball Brothers bought the plant.

The two routes to the ware class also converge. The Bottle Research Group reads a “3 RIVERS {star} HA” jar as a product or packers’ jar rather than a fruit jar, inferring the class from the mark [ev-0042]; the plant’s chemist states the class from the plant, as “a general line of packers’ ware” [@ev-0059, part 04, fol. 3430]. Neither knew of the other.

A preserved disagreement: what Smith wrote

The Bottle Research Group states that Smith called THREE RIVERS {star} the most common format, that Smith did not record the 3 RIVERS numeral variation at all, and that — since he “reported the ‘THREE RIVERS’ variation twice” — he probably intended one listing to be the numeral form [ev-0042].

Both witnesses of Smith’s page 33 held by this expert contradict that reading. The vision-OCR transcription of the original 1989 print scans and the 2024 re-typeset digital edition each read “The most common logo variation was ‘3 RIVERS,’ in all capitals, followed immediately by a five-pointed star,” with “THREE RIVERS” plus star listed as less common and THREE {star} RIVERS as rare; no variation is listed twice [@ev-0038, p. 33; @ev-0039, p. 33]. Smith’s price list cites a “3 Rivers {star}” side mark explicitly [@ev-0038, #52; @ev-0039, #52].

The two transcriptions are witnesses of one book and do not corroborate each other as independent sources would; what they establish is that the reading is not a transcription artefact of either process. The likeliest explanation is that the Bottle Research Group mischaracterized Smith’s text. The substantive conclusion is unaffected either way — 3 RIVERS {star} is the most common form both in Smith as transcribed here and in the Bottle Research Group’s observed sample — and the disagreement is recorded rather than resolved because the expert holds no third witness of Smith’s page 33.

The same chapter records a separate complaint: that Smith, with 94 photographed bottles and jars at his disposal, “did not make a more thorough examination of the examples” [ev-0042].

Works this article rests on but does not hold

Toulouse (1969, 1971), Giarde (1980), Roller (1983), and Creswick (1987) are named throughout above. None of them is admitted evidence here. Every claim attributed to them is a claim about what the Bottle Research Group reports them as saying [ev-0042], and a reader who needs one of those works settled must obtain it rather than rely on this article. The same applies to Lockhart’s Bottles on the Border El Paso observations, which reach this article through the same chapter [ev-0042].

The Hoffmann-Hayman catalogue

The Hoffmann-Hayman Coffee Co. of San Antonio maintains an active online catalogue of 75 numbered Three Rivers bottles plus an addendum — mostly sodas, plus the H&H “Crystalvac” coffee jars, dairy bottles, and an unknowns section. It identifies pieces by the Three Rivers star or “3R{star}” on base or heel and is keyed to Smith’s price list [ev-0006]. It is a confirming resource for the marks above rather than a source of new variants, and its value to the mission is that it is a working identification practice rather than a historical account.

The same catalogue documents the one mark on H&H ware that is not a Three Rivers mark: the Owens-Illinois Diamond-Oval-I, with plant code 7 for Alton, Illinois, on Crystalvac coffee jars dated 1936–1939 [ev-0006]. The star and the diamond both appear across the Crystalvac line, which is evidence that H&H took jars from Three Rivers and Owens-Illinois concurrently through the 1930s [ev-0006]. That is the single most consequential confusion this reference set exists to prevent.

What would settle the open questions

The variant date ranges are the weakest part of this reference. Nothing held here dates any variant more narrowly than 1925–1937 except the all-letters form, whose 1925–1927 range is stated as an educated guess [ev-0042]. A dated customer order, a mold-shop record, or a run of date-coded bottles across variants would settle it; none is held. The THREE {star} RIVERS variant rests on Smith’s report alone with no independent observation, and a single photographed example would move it.

The Mason-finish question added above opens a second want of a different kind. Hartford proposed in January 1935 to forbid a reorganized Three Rivers to make “jars with the Mason finish and seal” [@ev-0059, Gov. Ex. 203, part 16, fol. 59030]; whether any executed licence carried that clause would turn a possibility into a dating rule for every screw-finish Three Rivers jar. The executed agreements are not in the extract this expert holds, which covers four of the trial record’s fifty-eight parts. Neither is the document that would record the clause without the agreement: Hartford’s own ledger of licences and their amendments, printed in part 18, of which this expert holds two windows and nothing between them [@ev-0059, part 18, line 68242]. Of the two wants that is the cheaper, because it is a known document in a part already retrieved, and what is wanted is the pages between pages already held.