Who Made the Crystalvac Jars

Crystalvac was the Hoffmann-Hayman Coffee Company‘s vacuum-packed glass coffee line, launched in San Antonio in 1932. Its jars are among the most commonly surviving Three Rivers containers, and they are also the containers on which the question “is this a Three Rivers bottle?” is hardest to answer, because some Crystalvac jars carry the Owens-Illinois mark instead.

The two halves of that finding are not equally well supported, and this article keeps them apart.

The Three Rivers half is attested

The order is documented in the press as it happened. The San Antonio Register of June 10, 1932 ran “H and H Has New Container,” reporting that Hoffmann-Hayman, “one of the oldest coffee roasting concerns in the South, has announced a new vacuum packed container for their super-blend H and H brand: Crystalvac.” Gus J. Menger, the company’s president, is quoted on why: “to make it more possible for our customers to have the freshest of coffee… Each Crystalvac container is packed under 25 to 30 inches of vacuum.” The paper states plainly where the glass came from: “The initial order of one-quarter million Crystalvac containers was awarded the Three Rivers Glass company, a Texas concern, and was one of the largest individual glass orders ever given Texas” [@ev-0040, 1932 June 10 San Antonio Register]. The same notice records that the vacuum-packing plant meant twelve new hires on a force of fifty-six, and that H&H would keep selling coffee in cans as well [@ev-0040, 1932 June 10 San Antonio Register].

Forty-one years later, three separate Live Oak County collectors submitted inventories to Thelma Lindholm’s historical-marker application, each list headed as containers bearing the Three Rivers factory trade mark. All three hold Crystalvac jars:

• Hazel R. Gattis of George West: “2.. Square Crystalvac Coffee jar (Clear). 1.. [same] Jar (Amber)” [@ev-0041, p. 85].
• Mrs. R. Y. Williams of George West: “1.. Amber qt. H and H Crystaluae. On bottom, ‘Container Reg. U.S.Pat.Off. 601-5’” [@ev-0041, p. 86].
• Mr. and Mrs. Lloyd T. Cannon of George West: “2.. Clear H & H clear, quart square jar. 1.. Brown H & H square quart jar” [@ev-0041, p. 88].

Three things follow. The jars are square, and they come in clear and amber — the amber is not a misreading, it is in three independent hands in 1973. The base of the Williams jar carries “Container Reg. U.S. Pat. Off.” and the mould number 601-5, which independently confirms the 601-series mould codes the H&H collection reports on its Three Rivers jars [ev-0005]. And two of the three describe the jar as a quart.

Smith’s price list, entry #34, lists “H and H Coffee Crystalvac jars” among the Three Rivers customers — San Antonio, clear and amber, “We Roast It — Others Praise It” lid [ev-0006].

One caution about the 1973 lists: they were solicited as lists of Three Rivers containers. That they contain no Owens-Illinois-marked Crystalvac jar is what the request asked for and is not evidence that none existed.

The Owens-Illinois half rests on one holder

Everything this expert holds about Owens-Illinois-marked Crystalvac jars comes from handhcoffeefactory.com. The company’s own site reports that Crystalvac jars carry two marks — the Three Rivers star on the one-pound jars, and the Owens-Illinois Diamond-Oval-I on both one-pound and three-pound jars, in clear glass only — with plant digit “7” and single-digit year codes reading 1936 and 1938, plus one three-pound jar read as “2” for 1932. The site’s own conclusion is overlap rather than succession: “Collection specimens now date O-I supply to 1936–1939, overlapping the Three Rivers period rather than succeeding it,” and H&H was “almost certainly dual-sourcing through the 1930s.” It leaves the causal question open in its own words: “The unresolved mystery is whether Owens-Illinois replaced Three Rivers after it ‘was shutdown in 1937’” [ev-0005].

That is one source, and three of this article’s inputs are it. The full per-entry capture of the H&H bottle catalog [ev-0006] and the base photograph of the three-pound jar [ev-0020] are further pages of the same website, not further witnesses. The photograph in particular is not independent corroboration of the 1932 date code: it is the same holder’s image of the same jar the same holder already described.

Two custody facts belong with it. This expert does not hold the photograph — what is admitted is a written reading of it, and that reading’s own standing caveat is that it is “one weakly-struck mark read from a single photograph, not a bench examination of multiple examples” [ev-0020]. And the account of the site itself is a fetch synthesis assembled across five of its pages [ev-0005], so its quotations are quoted at one remove.

The literature that would settle this does not contain it. Lockhart and the Bottle Research Group’s Owens-Illinois work states the matter directly: “Hoffmann-Hayman / Crystalvac does not appear anywhere in this SHA/BRG corpus — its O-I attribution rests solely on the H&H collector site” [ev-0050].

What the plant code does settle

If a Crystalvac jar carries plant digit 7, it was not made in Texas. The complete Owens-Illinois plant roster — compiled by the Bottle Research Group with two former O-I employees against company records — gives plant 7 as Alton, Illinois, 1930–1983, and contains exactly one Texas location in the firm’s history: Waco, the second holder of plant number 15, from 1944 [ev-0054]. Waco opened February 22, 1944, on land bought in 1940; the nearest O-I facility to Texas in the 1930s was Okmulgee, Oklahoma, idle after 1932 [ev-0050].

This kills a durable piece of Three Rivers folklore in this particular application. Whatever undercut the Three Rivers plant, it was not a local Owens-Illinois factory, because there was none — the Palestine plant sometimes named in that story was Knox Glass, and it opened in 1941. Any O-I glass reaching a San Antonio coffee roaster in the 1930s came from Illinois by rail, concurrently with the Three Rivers orders rather than in place of them [ev-0050].

The dating rule is equally firm and is what makes the marks readable at all: on the Diamond-Oval-I mark the left number is the plant and the right digit is the year; a single undotted digit is a 1930s year, a dot after it marks 1940–45, and double digits begin in 1946 [ev-0050]. Three Rivers, by contrast, never used date codes — its jars are datable only by mould variant and the span of the mark [ev-0050].

What the catalog does and does not show

The first Owens-Illinois catalog, held here as two of its twenty-five scanned parts, shows that O-I sold vacuum coffee jars in the Crystalvac idiom. Its Coffee & Pantry Jars pages list nine moulds, including square waffle-grid jars with a round label panel — C-1011 and C-1012, the configuration closest to the Crystalvac form — all on the G-450 finish [@ev-0016, pp. F56-F57]. The Closure Division supplies the other half of the system: “Knife-Opening Vacuum Caps” lithographed “VACUUM PACKED — Keeps Contents Fresh,” offered in the same “G” GCA 450 thread, and pictured in commercial use on a branded coffee jar [@ev-0017, p. Clo-11]. The catalog even distinguishes vacuum from non-vacuum variants within the same thread family [@ev-0017, p. Clo-12].

What that establishes is capability, not a transaction. Nothing in the catalog names Hoffmann-Hayman or Crystalvac. A prior compiled account of this question held that the catalog closed the documentary case for dual sourcing from the start; it does not, and the distinction matters, because the whole weight of the O-I attribution still rests where it rested before.

Three further cautions on the catalog:

• Its date is not as tight as “1930–1935” suggests. The Historic Bottle site dates it from internal evidence and reaches two answers in the same discussion: that it “was put together over a period of several years beginning in 1930 and ending in 1935,” and, after a correspondent’s Donald Duck tumbler and the 1935 Libbey absorption, that “the catalog was published in 1935 at the earliest or 1936 at the latest” [ev-0018]. It is therefore weak evidence about what O-I was actively selling in 1932 — the year the contested jar is dated.
• Both catalog parts are vision transcriptions of image-only scans, made by an agent from PDFs with no text layer, and both are recorded as partial [@ev-0016; @ev-0017]. Twenty-three of the twenty-five parts are not held, and at least two of the unheld ones also contain coffee jars [ev-0018]. The nine moulds are the nine on those two pages, not O-I’s coffee line.
• The capacity arithmetic is not established. The transcription note reads the 28–30 fluid-ounce moulds as sitting “in roughly the capacity territory a 3-lb coffee jar implies” [ev-0016]. That is the transcriber’s inference, not the catalog’s, and it sits badly beside the 1973 inventories, which call the Three Rivers H&H jar a quart [@ev-0041, pp. 86, 88] while the H&H site assigns Three Rivers the one-pound size [ev-0005]. Nothing held gives the fluid capacity of the H&H three-pound jar, so nothing held connects it to a catalog mould.

A chronology problem the older account did not have

The H&H site’s chronology has Three Rivers “reorganized under Ball interests in December 1936” and the Texas corporation dissolved in January 1937, with Owens-Illinois supply running to 1939 [ev-0005]. The trial record in Hartford-Empire Co. v. United States does not read that way. The November 27, 1936 event was a trustee’s sale to Guardian Loan Company, confirmed in the District Court at George West; Ball had been the undisclosed prospective purchaser since at least September of that year; and George A. Ball told Hartford that “Three Rivers has not been operated since December” — December 1937, not 1936 [ev-0059].

This matters here because the site’s open question is framed on a 1937 shutdown. On the trial record the plant was still running through 1937, which widens rather than narrows the window in which both suppliers could have been filling H&H orders — and removes the tidy sequence in which O-I picks up after a 1936 collapse. The two chronologies are not reconciled by anything held.

The Toulouse claim, and a link to the fruit-jar dispute

Lee Troell, then a Live Oak County commissioner, wrote to the marker committee in 1973: “The Amber colored H and H jar is listed in the Book, ‘Fruit Jar’ by Toulouse. Toulouse designed this jar” [@ev-0041, p. 84]. The binder separately cites Toulouse’s Fruit Jars at pages 306 and 522 for the Three Rivers listing and the “3 Rivers ☆” mark [@ev-0041, p. 89].

This expert does not hold Toulouse, and the register of works it does not hold already carries Fruit Jars (1969) for other reasons. Troell’s claim is uncorroborated by anything held and is recorded here as a claim, not a fact. It is worth recording because Julian Harrison Toulouse was a glass engineer as well as a cataloguer, and the plant-number table that the Bottle Research Group rebuilt for Owens-Illinois began as his [ev-0054] — so a designer-level link between the amber H&H jar and the Owens-Illinois world is exactly the kind of thing that would matter, and exactly the kind of thing that must not be assumed from one sentence.

The Crystalvac jar also lands in the middle of the dispute over what Three Rivers was permitted to make. Its threading fit standard Kerr and Mason-Ball lids, and housewives reused the jars for home canning [ev-0005]; the San Antonio Register advertised exactly that use — “use the glass container over and over again for reserves” [@ev-0040, 1932 June 10 San Antonio Register]; and a 1973 informant places the jar in a book called Fruit Jars [@ev-0041, p. 84]. Against that, the plant’s chief chemist swore that Three Rivers made “fruit-packers ware, one trip container; they weren’t Mason type jars,” adding that “I think Mr. Ball said they did” sell them to be used as fruit jars [ev-0059]. The Crystalvac jar is a concrete case of the ambiguity that dispute turns on: a reusable, Mason-threaded container sold as a coffee package.

Where this leaves the identification question

• A square clear or amber quart Crystalvac jar bearing the 3 Rivers star, with a 601-series mould number, is a Three Rivers jar. That is attested [@ev-0040; @ev-0041; @ev-0006; @ev-0005].
• A Crystalvac jar bearing the Diamond-Oval-I was not made in Texas; read the plant digit and the date digit, and remember that an O-I mark does not imply a post-1940 jar [@ev-0005; @ev-0050; @ev-0054].
• That Owens-Illinois supplied Hoffmann-Hayman at all is reported by one holder and appears in no independent source this expert holds. It is plausible, it is specific, and it is uncorroborated [@ev-0005; @ev-0050].
• Dual sourcing is the best available reading, not an established fact. The documentary case for it is a capability argument [@ev-0016; @ev-0017], the physical case is one holder’s jars [@ev-0005; @ev-0020], and the circumstantial case — that Three Rivers ran H&H glass as made-to-order batches rather than stock, since “container/bottle factories did not keep large stocks of bottles to sell… They would not keep on hand large number of branded bottles, for H&H, Trapeys, etc.” [ev-0019] — explains why a second supplier would be convenient without showing that there was one.

What would settle it

An Owens-Illinois-marked Crystalvac jar in a hand other than Hoffmann-Hayman’s, or a photograph of one admitted as an image rather than as a description of an image. Failing that, the remaining twenty-three catalog parts, Toulouse’s Fruit Jars, or any H&H purchase record.