Three Rivers Glass — Manufacturing Process

The factory published an illustrated tour brochure that walks its line step by step — batch, furnace, forming machines, fire-finish, lehr, inspection — and it survives in Thelma Lindholm’s marker research [ev-0041]. This article follows that brochure and names the standard technology of the period behind each step, so that its plain-English captions can be read as the industry practice they describe.

The mature process, after mechanization in 1925, was the ordinary automatic soda-lime-silica bottle process of its era [ev-0003], with two local advantages the company marketed hard: its own Carrizo quartzose sand and abundant natural gas [ev-0041]. It had not always made glass this way, and the evolution is at the end.

A caution about the brochure. It is the company describing itself to visitors and customers. Its process sequence matches independent accounts of how bottle houses of the period worked [@ev-0003; @ev-0008], which is good reason to trust its shape; its numbers are its own and are corroborated by nothing held here.

1. Batch

The recipe is soda-lime-silica, the standard bottle glass: silica sand as the former and the bulk of the mix, soda ash as the flux — pure silica melts near 1700 °C and soda brings the working range down to a practical 1400–1600 °C — and lime as the stabilizer that makes the glass chemically durable rather than water-soluble [ev-0003]. Three Rivers used local Carrizo quartzose sand [ev-0041].

Cullet, recycled glass, was used by virtually every plant to speed melting [ev-0003]; the brochure does not mention it, and nothing here says whether Three Rivers used it.

In the brochure’s own words, “exact proportions are first weighed and dumped into the hopper … elevated and thrown into the cylinder-mixer which revolves at high speed. From here it is fed into the furnace” [ev-0041]. That is the textbook batch-house sequence — weigh, mix for homogeneity, charge the furnace [ev-0003].

2. Melting

The plant ran a continuous tank furnace of 125-ton capacity [ev-0041], the industry-standard technology by 1922; older pot and day tanks had been abandoned by bottle houses a decade earlier [ev-0003]. Batch is charged at one end, melts, and flows continuously toward the working end, and the furnace burns through regenerative brick checkers that preheat the combustion air [ev-0003]. Three Rivers burned natural gas [ev-0041].

A tank furnace runs around the clock, because reheating a tank from cold is both wasteful and damaging to the refractory [ev-0003]. The brochure’s “45 tons of molten glass taken daily” is therefore a steady-state draw and not a shift figure [ev-0041].

Between melting and forming the glass is fined, so bubbles rise out, and conditioned, so it cools to working viscosity [ev-0003]. That is what accounts for the brochure’s “3½ days from raw material to finished bottle” [ev-0041].

3. Forming

After 1925 the ware was formed on Lynch rotary machines fed by Hartford-Empire gob feeders. The brochure names “a gigantic Lynch machine and Hartford-Empire feeder … Twenty-five thousand beverage bottles every 24 hours … we have … THREE of these,” plus a separate rotary “giant milk-bottle machine” [ev-0041].

The gob feeder. Molten glass flows into the feeder; a plunger forces it through an orifice and a mechanical shear cuts the stream into a precise, timed gob. This feed-and-flow method is a Hartford-Fairmont and later Hartford-Empire design, essentially unchanged since 1918, and it is what let a rotary machine run fully automatic. Unlike the rival Owens machine it leaves no suction scar [ev-0008].

Blow-and-blow, for narrow-neck beverage bottles. The gob drops into a two-piece blank or parison mould; compressed air first blows it down to form the finish, the sealing mouth — the settle blow — and a counter-blow forms the rest of the parison; transfer arms invert it into the blow mould, whose walls carry the maker’s mark and any customer embossing; a final blow shapes it, the mould opens, and a takeout carries it off [ev-0008]. The brochure’s timing is “Twenty seconds after a gob … is dropped into the molding machine, a bottle or jar emerges” [ev-0041].

Press-and-blow, for wide-mouth ware. In press-and-blow a plunger presses the gob into the blank mould to form the parison rather than blowing it, and the parison is then blown to final shape; this is how wide-mouth milk bottles and jars were made [ev-0008]. The brochure’s separate “milk-bottle machine” is described as a “circular table-like base … revolves; stopping momentarily at each mould until an exact quantity of molten glass can be deposited,” after which “compressed air … blows the glass automatically against the interior sides of the moulds” [ev-0041].

The machine has a name in the record: an LA machine. The brochure never says what it was, and for as long as the brochure was the only account of this floor, it could not be said. The trial record says it in passing, in a cross-examination about somebody else’s résumé. Steve Coleman, the plant’s chemist, was asked what his later business partner had done at Three Rivers:

He operated a milk bottle machine, an LA machine, there for quite a period of time when he first went to work at Three Rivers. [@ev-0059, part 04, fol. 3452]

That is the only model designation this expert holds for any forming machine at this plant, and it belongs to the one machine the brochure singles out. The man who ran it is named too: H. O. Knape, who went from this machine into the mould shop and was for some time its foreman [@ev-0059, part 04, fol. 3452] — followed further in Who Cut the Molds.

“LA” is a Lynch type, on the same witness’s own usage. Coleman does not say Lynch at fol. 3452. He says it earlier in the same testimony, on direct, listing what he and Knape found at the Santa Anna plant they bought in 1933: “We had two Lynch LA type machines, and one Miller milk bottle machine, and there were two Miller feeders” [@ev-0059, part 04, fol. 3431]. So “LA” is a maker’s type designation in this witness’s vocabulary and the maker is Lynch — the same maker the brochure names for the beverage machines [ev-0041], and a firm that designated its models in exactly this way: the Knox plant at Palestine ran “four Lynch 10 machines, a Lynch JP machine, and a Lynch JPM machine” [ev-0052].

Two cautions, because a sentence about a different plant is doing the work. Coleman did not say “Lynch” about the Three Rivers machine; the identification is his usage carried across twenty-one folios, not his statement. And at Santa Anna the milk-bottle machine was a Miller, while the LAs were the other two — so “LA machine” is not his shorthand for “milk-bottle machine.” It is a machine type that happened to be running milks at Three Rivers and something else at Santa Anna, which is what a general-purpose rotary of the period was for.

That it worked press-and-blow is still an inference, and the name does not close it. The brochure describes a rotary wide-mouth machine and does not say how the parison was formed; press-and-blow is what the period used for wide-mouth ware [ev-0008]. A model name is not a mechanism, and nothing held describes a Lynch LA — no cycle, no station count, no drawing. What is held nearby is a Lynch Milk Bottle machine documented with a seven-stage press-and-blow cycle [ev-0008], and nothing held says the LA is that machine or that it is not. The inference is better placed than it was and is still an inference.

A sidelight that points the same way. A trade letter of 4 May 1933 among the government’s exhibits reports that “the Hartford-Empire people has bought the Miller Machine Company, they made principly press and blow type of machines” [@ev-0059, part 16, fol. 60429]. The Santa Anna plant’s milk-bottle machine was a Miller [@ev-0059, part 04, fol. 3431]. That is period trade usage tying the milk-bottle machine of Coleman’s own world to press-and-blow. It is a sidelight and not a proof: a different maker at a different plant, saying nothing about the Lynch LA.

And it was not on a Hartford feeder. The brochure presents the milk-bottle machine as simply a different machine for a different bottle. It was also the one piece of the line running outside the company’s licence, by an arrangement Hartford insisted on and wrote down; that is set out under Evolution below.

Not the Owens machine. The Owens suction machine, from 1905, needed no feeder: it sucked glass straight from the tank and left a diagnostic suction scar. The cheaper gob-feeder-and-rotary approach displaced it. The Individual Section machine, from 1925, later superseded the rotary Lynch table across the industry [ev-0008]. Three Rivers ran the older rotary architecture [ev-0041], which matters for identification: a Three Rivers bottle should carry no suction scar.

4. Fire-finishing

Every forming machine leaves a sharp edge at the mouth, and fire-polishing — reheating the formed rim — is the standard mechanized step for smoothing it. It is distinct from the separately tooled applied finish of the earlier mouth-blown era [ev-0003]. On the Three Rivers conveyor, bottles “still red hot … are brought to a short stop directly under each blue-flame gas blast … FIVE of these jets, burning natural gas,” melting “the sharp edge inside the finish ring” [ev-0041].

5. Annealing

Formed glass cools unevenly — the skin hardens while the core is still hot — which locks in internal stress that would make a bottle crack spontaneously. A lehr, a temperature-zoned tunnel oven on a moving belt, holds the glass near the annealing point and then cools it slowly and evenly to relieve that stress [ev-0003]. At Three Rivers the ware travelled roughly two hours through the lehr, loaded by a “Hartford-Empire stacker” whose arm “grasps each piece and lifts it into the lehr” [ev-0041].

The brochure glosses annealing as meaning “in reality … tempered” [ev-0041]. That is a period lay-misnomer and worth correcting, because the two are opposites: annealing relieves internal stress, which is what the brochure otherwise describes correctly, while tempering deliberately adds a stress profile to strengthen glass [ev-0003]. The process is annealing; only the gloss is loose.

6. Inspection and testing

“Expert packers inspect each one as they come off. Bottles having any defects are immediately discarded” [ev-0041]. Beyond visual sorting the plant ran a laboratory battery, and every test in it is aimed at residual stress or weakness left by forming and annealing: a thermal-shock test by immersion in 146 °F water, a finish and wedge-expansion test inside the neck, a pressure test in which bottles “stand over 600 pounds pressure” against roughly 90 lb of actual capping force, and the polariscope [ev-0041].

The polariscope uses polarized light to make residual stress visible as coloured fringes — the brochure’s “prismatic reaction” — and it is the standard non-destructive check that the lehr did its job [@ev-0003; @ev-0041].

Evolution: hand-blown, then Tremblay’s machine, then a licence with a hole in it

1922 to 1924, hand blowing. Production began hand-blown, with skilled glassblowers brought in from the north, and no pre-1925 ware is marked [ev-0042]. A hand shop gathered glass on a blowpipe, blew it into a mould, and hand-tooled a separate applied finish [ev-0003].

1924 to 1925, Tremblay’s machine. Facing the Hartford-Empire Company’s near-total patent monopoly on automatic machinery, Tips hired the engineer Albert F. Tremblay to design a bottle-forming machine of the company’s own. It is documented in U.S. Patent 1,648,792, “Bottle-forming machine,” filed 26 June 1926 and granted 8 November 1927 — a compact rotary machine with blank and finishing mould stations [ev-0051]. The company ran automated equipment from about early 1925 [ev-0042], which is before the patent issued; the machine is therefore best described as later-patented rather than as patented technology the plant adopted.

Whether it was also a workaround — a design that got around Hartford’s patents rather than into them — is the question the collector literature answers easily and the trial record answers the other way.

1929, an infringement notice, a concession, and then a licence. This article’s predecessor had Hartford extending a licence in August 1929 “after several years of proven output” [ev-0042], as though the machine had earned it. Tips’ own letter of 27 May 1929, to the chairman of the Thatcher Manufacturing Company and over his signature as general manager, says what actually happened:

“As I explained to you, the Hartford Empire [Company has] notified us that the feeders we are now using infringe their patent rights. We have not contested this claim, but have agreed with them on a payment of release of damages.” [@ev-0059, part 16, fol. 60713]

So the sequence is notice, concession, payment, licence — not proven output rewarded. The Three Rivers feeders were not outside Hartford’s patents, and the company did not argue that they were.

The licence was granted with a hole left in it, deliberately. Hartford would not grant Three Rivers milk-bottle rights: those licences had been assigned to the Thatcher Manufacturing Company in 1920 [@ev-0059, part 09, fols. 10273-10274], and Thatcher had told Tips in October 1928 that extending them would “get into trouble with other licensees” [@ev-0059, part 16, fol. 60711]. Three Rivers was already making milk bottles. Rather than lose that trade, Tips took a licence covering everything else and kept one home-built feeder running on milks, outside it. Hartford’s own account of the bargain, written to Tips on 1 June 1931:

“As you will remember, we hesitated for a considerable period of time before offering you a Hartford License, in view of the fact that you definitely stated you felt you must continue to operate one of your ‘Three Rivers Feeders’, in view of the fact that Hartford was unable to grant you Milk Bottle rights. However, we did offer you a license for certain types of ware other than Milk Bottles, with the knowledge that you would continue to operate one feeder on milk bottles, which feeder would not be under license from Hartford.” [@ev-0059, part 16, fol. 60862]

“This license you accepted with the understanding that Hartford would, at some time dictated by its own judgment, bring suit against the Three Rivers Glass Company on this one unlicensed feeder, which we considered to be an infringement of Hartford patent rights. This situation, of course, remains unchanged at the present time.” [@ev-0059, part 16, fol. 60863]

The same letter records the other half of the bargain: the unlicensed feeder “would be used solely for the manufacture of milk bottles,” and this was “one of the definite conditions under which we granted and you accepted a license” [@ev-0059, part 16, fol. 60863]. What Hartford was licensing was not a machine but a Field of Ware, and it policed the boundary — the 1931 letter is itself a complaint that a pressed packers tumbler had appeared in the trade from “a licensee new in the pressed business,” which “causes us considerable embarrassment” [@ev-0059, part 16, fol. 60862].

The unlicensed feeder was frozen in place, and inspected. Hartford’s president, writing privately in December 1929: “As far as we know, Three Rivers are operating their own home-made feeder on milks. They have been told in writing, and have agreed, that they must not change this feeder in any way, shape or manner, and we, of course, have refused to license them on this feeder, or to grant them a license for any Hartford equipment on milks. They are building up infringement damages, and if we win the decision in the Court of Appeals in St. Louis, we will be in a position to get an injunction” [@ev-0059, part 16, at fol. 60719]. He adds that on the next call at the plant “we will make a special point of seeing whether they have done anything to their own feeder” [@ev-0059, part 16, at fol. 60719].

This is the single most useful technical fact the trial record supplies about the plant. From 1929 the milk-bottle feeder could not be developed. Whatever improvements the licensed side of the shop took from Hartford, the machine making milk bottles was contractually held at its 1929 configuration for the rest of the company’s life, under periodic inspection by the patent holder.

The arrangement outlived the company that made it. Asked under oath in the antitrust trial whether “Three Rivers operated one feeder on milks free during this time,” Hartford’s Safford accepted the premise and corrected the word: “By ‘free’ you mean in defiance of our patent rights. We regarded it as an infringing feeder.” He was then asked whether Hartford knew, when it licensed the receiver, that the receiver would go on running it: “Yes, that is correct; and I think that he realized that, too” [@ev-0059, part 09, fols. 10273-10274]. The receivership inherited the hole in the licence along with the plant.

What this means for reading the ware. From roughly 1929 the plant ran two classes of forming equipment side by side: Lynch machines on licensed Hartford-Empire feeders for the licensed field of ware, and one “Three Rivers Feeder” — Hartford’s own term for it, and Goodwin Smith’s “home-made feeder” — on milk bottles alone. The brochure’s separate “giant milk-bottle machine” is that division, described from the visitor’s side without the reason. Anyone comparing a Three Rivers milk bottle with the rest of its output is comparing ware off two different machine lineages, one of which was legally forbidden to change.

What this account rests on, and what would improve it

The shape of the process is secure: it matches independent descriptions of how bottle houses worked in the period [@ev-0003; @ev-0008], and its most specific elements — the Hartford-Empire feeders and the Tremblay machine — are corroborated outside the brochure [@ev-0042; @ev-0051]. The licensing arrangement behind the forming floor is now the best-evidenced thing in this article: it rests on the parties’ own letters and on sworn testimony, produced under subpoena in the federal antitrust case [ev-0059].

The numbers are not. The 125-ton tank, the 45 tons drawn daily, the 25,000 bottles per machine per day, the three machines, the twenty-second cycle, the five gas jets, the two-hour lehr, the 146 °F and 600 lb tests: every one of these reaches this expert through a single promotional document [ev-0041]. An insurance survey, a state factory inspection, an equipment invoice, or a trade press account of the plant would corroborate or correct them, and none is held.

What the trial record supplies is mostly the legal shape of the machinery rather than its engineering. It gives no dimension, no cycle time and no output figure for either class of feeder, and it never describes how the “Three Rivers Feeder” actually worked — only that Hartford considered it an infringement and forbade its alteration. Tremblay’s patent [ev-0051] remains the sole technical description this expert holds of any machine built at this plant, and the patent is of a forming machine, not of the feeder that fed it. Whether the “Three Rivers Feeder” is the feeding mechanism of the patented machine, a separate device, or a modified Hartford unit is not settled by anything held here.

It supplies exactly one piece of hardware nomenclature, and the gap that opens behind it is the sharpest acquisition want in this article. The milk-bottle machine was an LA [@ev-0059, part 04, fol. 3452]; no document held describes what a Lynch LA was. A Lynch Glass Machinery Co. catalogue, an equipment invoice, or a trade-press notice of the model would turn a name into a machine and would settle press-and-blow for the one machine at this plant whose product collectors handle most. The plant’s feeders were separately numbered by the licensor — Hartford’s own records name a “Single Feeder No. 295” here — and that is a serial, not a model, and it identifies nothing about how the machine worked [@ev-0059, part 16, fols. 60789-60791].