references · updated 2026-08-28
Three Rivers Glass — Known Customers
confidence: attested weakest: ev-0019
Roster of roughly sixty known customers identified by embossed names on surviving marked bottles and jars
Who bought Three Rivers glass? The evidence is the bottles. When a piece carries both a Three Rivers maker’s mark and an embossed customer name, it documents at least one order. By that method the record names roughly sixty customers: Texas soda franchises and independent bottling works above all, plus dairies, food packers, and a few out-of-state buyers from Georgia to New Mexico.
The list is a floor and not a census, and the rest of this article is mostly about why.
How to read this list
Every row here exists because somebody paid for a name in the glass, and the trade had a reason to pay. Container factories “did not keep large stocks of bottles to sell”; Three Rivers “sold glass, payment on delivery,” and was out the cost of gas, sand and labor until the load moved [ev-0019]. What the plant did keep on hand was “small runs of generic soda bottles, for someone that could not afford to order in quantity to pay for their town or company name on the bottle.” Most buyers paid for the name, because “the bottles were a large investment,” refilling another man’s bottle was illegal, and the embossing was how a bottler proved which bottles were his [ev-0019]. So the roster below is a list of customers who ordered in quantity. It has never been a list of customers, and the reason is commercial rather than archival.
Six caveats govern every entry.
Brand is not purchaser. On franchise brands the customer was the local franchise plant, not the brand owner; Coca-Cola mandated city embossing in 1918 precisely so that local plants could be told apart [ev-0046].
“Made for” is not necessarily “ordered by.” Plate-mould embossing identifies the party a run was manufactured for, and jobbers or brand licensors sometimes ordered on a bottler’s behalf [ev-0046].
Manufacture is not use. Bottles were reused for years and refilled by competitors despite “property of” and “never sold” embossing, so where a bottle was found adds nothing to the order book [ev-0048].
Date codes are unsettled. One Lockhart chapter states that Three Rivers never used date codes [ev-0045], yet his own documented bottles carry a 1928 code on a Mistletoe half-pint [@ev-0045; @ev-0058] and a 1932 code on a Dairy League half-pint [ev-0045], and codes from 1928 are reported independently [ev-0042]. This is a live conflict in the sources and it is preserved here rather than resolved: treat an individual code as probable, not certain.
The list systematically undercounts. Generic ware — plain sodas, packers’ jars, fruit jars, plain milk bottles — carried no customer name at all, and Smith’s own list contains such entries [@ev-0038; @ev-0039]. Absence from this list is not absence of a relationship. The trial record puts a size on that: a statewide wholesale-grocery trade in unembossed jars, complained of by a competitor as large enough to threaten its fruit-jar business [@ev-0059, Gov. Ex. 948, part 16, fol. 60285], of which this roster shows nothing. See The trade the bottles cannot show, below.
One bottle is one order. Small bottlers reordered every year or two in minimum lots and switched glasshouses freely, so an entry documents at least one order and not loyalty [ev-0048].
Evidence tiers
Rows below are marked by what stands behind them. [S] is Smith’s 1989 price list, held here in two independent transcriptions of the same book [@ev-0038; @ev-0039]. [P] is a Smith photo plate — a photographed specimen [ev-0038]. [L] is a Lockhart-documented specimen, physically examined and published, and is the strongest tier. [H] is the Hoffmann-Hayman collector catalogue [ev-0006]. [B] is the Bottle Research Group chapter [ev-0042]. [N] is a period newspaper [ev-0026]. [T] is the Hartford-Empire trial record [ev-0059], which embosses nothing and names no customer by mark; it is used here where it documents a trade the embossed method cannot see, and once where it corroborates a ware class two of the rows below rest on.
The photo plates were mined caption by caption against the price list. They yield exactly two customers the price list lacks — Jay Bee Liquid Bluing, embossed, and Crown Brand dill pickles, paper-labelled. Every other captioned name already appears below [ev-0038].
El Paso is exhausted
Lockhart’s complete El Paso corpus — all 23 hosted Bottles on the Border chapter files and all 20 El Paso Dairies files — was searched by a stated and reproducible method. No Three Rivers customer appears anywhere in it beyond the seven already listed, and one negative case [ev-0058].
That is a measured result rather than an absence of effort, and it carries a second point. El Paso is the most thoroughly documented bottling market in Texas and it sits about 600 miles from the plant, so a footprint of seven customers there is thin. That thinness is itself evidence for reading the customer base as a truck-distribution radius [ev-0058] — an inference the sweep’s own author draws, and one no other source held here tests.
The trade the bottles cannot show
Every row below rests on an embossed customer name. The trial record documents a Three Rivers trade in ware that carried none, and documents it through a competitor with every reason to measure it accurately.
In March 1933 an interoffice letter out of Dallas, addressed to Hazel-Atlas Glass Co. at Wheeling 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,” 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]. Two months later a vice-president and general sales manager wrote to his own president that “These people are very active in Texas, going to the wholesale grocery trade all over that State and selling plain pint and quart jars, without caps,” at prices which, with two-piece Bernardin or Crown caps added, ran “so much lower than the Ball list on fruit jars as to make the situation almost impossible for the legitimate fruit jar manufacturer” [@ev-0059, Gov. Ex. 948, part 16, fol. 60285]. The plant’s chief chemist described the product line the same way under oath: “a general line of packers’ ware, and for several years manufactured milk bottles” [@ev-0059, part 04, fol. 3430].
Not one jobber, wholesale grocer, or packer in that trade can appear in the tables below, because a plain jar sold without a cap carries no customer’s name. The fifth caveat above now has a size. A competitor’s sales management thought this trade large enough to be worth stopping through the licensor, and this roster shows none of it.
It also qualifies the truck-radius reading. The El Paso sweep found seven customers 600 miles out and read the thinness as a distribution radius [ev-0058]. Sherman is in the far north of Texas and “all over that State” is the competitor’s own phrase, so the plant’s ware plainly travelled further than its bottles suggest. The two readings are not in conflict once the ware classes are separated: sodas and milk bottles were returnable, heavy, and refilled locally, so their trade really is a radius; cased jars sold through jobbers went one way on freight and never came back — “one trip container,” in the maker’s own words [@ev-0059, part 04, fol. 3430]. The radius is a property of the returnable-bottle trade, not of the company. Nothing held measures either reach directly, and both readings remain inferences.
A third reach is claimed, and this roster can test half of it. Mackey’s account has the company running a general sales office in Dallas and district offices in Houston, Oklahoma City, Little Rock and New Orleans, with glass going “as far as Georgia, Puerto Rico and Central America” [@ev-0040, p. 4], and describes the plant as having made “a small and rural South Texas town, an international concern” [@ev-0040, p. 2]. Georgia holds up: two rows below are Georgia customers, Brandimist of Atlanta [@ev-0038, #7; @ev-0039, #7] and a Columbus variant of Nehi [@ev-0038, #49; @ev-0039, #49], both embossed and both in Smith’s price list. Puerto Rico and Central America hold up nowhere. No such customer appears in the price list [@ev-0038; @ev-0039], none appears in the El Paso sweep [ev-0058], and the competitor correspondence that measured this company’s trade most closely measured it only in Texas [@ev-0059, Gov. Ex. 948, part 16, fol. 60285]. Nothing held corroborates the export claim or contradicts it, and an unembossed export trade would look exactly like this.
Franchise soda customers
The customer is the local franchise plant [ev-0046].
| Franchise plant | Location | Evidence |
|---|---|---|
| Coca-Cola Bottling Co. | Bastrop, TX | [S] #11 [@ev-0038, #11; @ev-0039, #11] |
| Coca-Cola “Gills” | Beeville, TX | [S] #12 [@ev-0038, #12; @ev-0039, #12] |
| Giddings Coca-Cola Bottling Co. | Giddings, TX | [S] #32 [@ev-0038, #32; @ev-0039, #32] |
| Coca-Cola Bottling Co., as bottler of Anderson’s Soda Water | — | [S] #1 [@ev-0038, #1; @ev-0039, #1], [H] [ev-0006] |
| Coca Cola Bottling Co. (Staat’s Beverages) | New Braunfels, TX | [S] #65 [@ev-0038, #65; @ev-0039, #65], [H] [ev-0006] |
| Dr. Pepper Bottling Co. | Eagle Lake, TX | [S] #18 [@ev-0038, #18; @ev-0039, #18] |
| Dr. Pepper Bottling Co. | Laredo, TX | [S] #20 [@ev-0038, #20; @ev-0039, #20] |
| Dr. Pepper Bottling Co. | Mt. Pleasant, TX | [S] #21 [@ev-0038, #21; @ev-0039, #21] |
| Dr. Pepper Bottling Co. | Paris, TX | [S] #22 [@ev-0038, #22; @ev-0039, #22], [L] near-twin of the Union El Paso bottle [ev-0049] |
| Dr. Pepper | Temple, TX | [S] #23 [@ev-0038, #23; @ev-0039, #23] |
| “DP Soda — Dr. Pepper” | Uvalde, TX | [S] #17 [@ev-0038, #17; @ev-0039, #17] |
| Empire Products Corp., Dr Pepper franchise | El Paso, TX | [L] clock-face bottle, ca. 1930 [ev-0046] |
| Pepsi Cola | Austin, TX | [S] #54 [@ev-0038, #54; @ev-0039, #54] |
| Orange Crush | Kenedy and New Braunfels, TX | [S] #53 [@ev-0038, #53; @ev-0039, #53], [H] variants [ev-0006] |
| Nehi Beverages | Columbus, GA variant | [S] #49 [@ev-0038, #49; @ev-0039, #49], [H] [ev-0006] |
| Royal Crown Root Beer | Morgan City, LA | [S] #59 [@ev-0038, #59; @ev-0039, #59] |
| Delaware Punch | — | [S] #15 [@ev-0038, #15; @ev-0039, #15] |
| Kist | — | [S] #41 [@ev-0038, #41; @ev-0039, #41] |
Independent bottling works and beverage makers
| Customer | Location | Evidence |
|---|---|---|
| Austin Bottling Works | Austin, TX | [S] #2 [@ev-0038, #2; @ev-0039, #2] |
| Big Chief — Blackburn Bros. | Clarksville, TX | [S] #5 [@ev-0038, #5; @ev-0039, #5] |
| Bluebonnet Beverages | Rusk, TX | [S] #6 [@ev-0038, #6; @ev-0039, #6] |
| Brandimist | Atlanta, GA | [S] #7 [@ev-0038, #7; @ev-0039, #7] |
| Burns | — | [S] #8 [@ev-0038, #8; @ev-0039, #8] |
| California | Dallas, TX | [S] #9 [@ev-0038, #9; @ev-0039, #9] |
| Carmine Bottling Works | — | [S] #10 [@ev-0038, #10; @ev-0039, #10] |
| “Contents 6½ Fl. Oz.” | Uvalde, TX | [S] #13 [@ev-0038, #13; @ev-0039, #13] |
| Crown Bottling Co. | West, TX (Smith’s own question mark) | [S] #14 [@ev-0038, #14; @ev-0039, #14] |
| Diamond K Beverages | Kingsville, TX | [S] #16 [@ev-0038, #16; @ev-0039, #16] |
| Dragon Bottling Co. | San Antonio, TX | [S] #24 [@ev-0038, #24; @ev-0039, #24] |
| Edward’s | — | [S] #25 [@ev-0038, #25; @ev-0039, #25] |
| F. J. Piwetz Bottling Wks. | Fayetteville, TX | [S] #27–28, two styles [@ev-0038, #27-28; @ev-0039, #27-28] |
| Fizzon (“Joy-Synth”) | — | [S] #29 [@ev-0038, #29; @ev-0039, #29] |
| The Famous Milwaukee Drinks | — | [S] #31, #47 [@ev-0038, #31, #47; @ev-0039, #31, #47] |
| H. Cuellar | Zapata, TX | [S] #35 [@ev-0038, #35; @ev-0039, #35] |
| Holt Beverage Co. | Waco, TX | [S] #33 [@ev-0038, #33; @ev-0039, #33] |
| Hondo Bottling Works | Hondo, TX | [S] #36 [@ev-0038, #36; @ev-0039, #36] |
| Ideal Bottling Co. | — | [S] #37, noted “very unusual” [@ev-0038, #37; @ev-0039, #37], [L] dual 3 Rivers/Ball, 1937 [ev-0042] |
| “It — The Whole Tired World Wants It” | San Antonio, TX | [S] #38 [@ev-0038, #38; @ev-0039, #38] |
| Julep | Pearsall, TX | [S] #39 [@ev-0038, #39; @ev-0039, #39] |
| Jumbo | — (“SBC,” possibly Southern Bottling Co.) | [S] #40 [@ev-0038, #40; @ev-0039, #40], [H] hedged [ev-0006] |
| Klasek’s Bottling Works | — | [S] #42 [@ev-0038, #42; @ev-0039, #42] |
| Knightcap | — | [S] #44 [@ev-0038, #44; @ev-0039, #44] |
| Kraus Bottled Purity | Fredericksburg, TX | [S] #43 [@ev-0038, #43; @ev-0039, #43] |
| Mexi-Pep | — | [S] #45 [@ev-0038, #45; @ev-0039, #45] |
| Mistletoe (soda) | — | [S] #48 [@ev-0038, #48; @ev-0039, #48] |
| Nicholson Bottling Works | El Paso, TX | [S] #50 [@ev-0038, #50; @ev-0039, #50], [L] ca. 1934 order; the 1937 dual-marked half-pint is Nicholson’s [ev-0047] |
| Pickoff “Sign of Purity” | Taylor, TX | [S] #55 [@ev-0038, #55; @ev-0039, #55] |
| Reed Bros. | Luling, TX | [S] #56 [@ev-0038, #56; @ev-0039, #56] |
| Rodriguez Root Beer Bottling Co. | San Antonio, TX | [S] #51–52 [@ev-0038, #51-52; @ev-0039, #51-52] |
| Rio Rey / Sharry Products Co. | Mission, TX | [S] #58, #61–62 [@ev-0038, #58, #61-62; @ev-0039, #58, #61-62] |
| San Benito Bottling Works | San Benito, TX | [S] #60 [@ev-0038, #60; @ev-0039, #60] |
| Southern Bottling Co. | Corpus Christi, TX | [S] #63 [@ev-0038, #63; @ev-0039, #63] |
| Southwest Ice Co. | Georgetown, TX | [S] #64 [@ev-0038, #64; @ev-0039, #64] |
| Standard Bottling Works (”…Never Sold”) | San Antonio, TX | [S] #66 [@ev-0038, #66; @ev-0039, #66] |
| Taylor Bottling Works | Taylor, TX (Smith’s own question mark) | [S] #67 [@ev-0038, #67; @ev-0039, #67] |
| “Texas” | Waco, TX | [S] #68 [@ev-0038, #68; @ev-0039, #68] |
| Threemor | Temple, TX | [S] #69 [@ev-0038, #69; @ev-0039, #69] |
| Trinity Bottling Works | Trinity, TX | [S] #72 [@ev-0038, #72; @ev-0039, #72] |
| Union B. Works | Houston, TX | [S] #73 [@ev-0038, #73; @ev-0039, #73] |
| Union Bottling Works | El Paso, TX | [L] two specimens, ca. 1925–1930 and 1930s [ev-0049] |
| Victory Bottling Works (Alejandro Marquez) | El Paso, TX | [S] #74 [@ev-0038, #74; @ev-0039, #74], [L] 1927–1931 [ev-0048] |
| Lone Star Bottling Works (Jose Urrutia) | El Paso, TX | [L] Ram Davila collection specimen [ev-0048] |
| Crystal Beverage Co. (Thomas O’Conor) | Alamogordo, NM | [L] two specimens, base “6”, probably 1936 [ev-0044] |
| UNEEDA Bottling Works | Beaumont, TX | [H] addendum specimen with star heel mark [ev-0006] |
| Yoakum C.V.C. Co. | Yoakum, TX | [S] #75 [@ev-0038, #75; @ev-0039, #75] |
Dairies and creameries
| Customer | Location | Evidence |
|---|---|---|
| Mistletoe Creameries | Fort Worth HQ, branches incl. El Paso, Dallas, Houston, Amarillo, San Antonio | [S] #46 quart and half-pint [@ev-0038, #46; @ev-0039, #46], [L] specimen with 1928 date code; the firm ceased in 1932 [@ev-0045; @ev-0058] |
| Lone Star Creamery | Houston, TX | [S] #46, pint [@ev-0038, #46; @ev-0039, #46] |
| Phoenix Dairy (“Milk — The Ideal Food”) | possibly El Paso | [S] #46, quart [@ev-0038, #46; @ev-0039, #46] |
| Roselawn Farms | Three Rivers, TX — the hometown dairy | [S] #46 [@ev-0038, #46; @ev-0039, #46] |
| Dairy League (Elephant Butte Dairy League) | El Paso, TX, 1931–1933 | [L] half-pint, 1932 date code [ev-0045] |
Food, condiment, and non-beverage customers
| Customer | Location | Product | Evidence |
|---|---|---|---|
| Hoffmann-Hayman (“H and H”) Coffee Co. | San Antonio, TX | Crystalvac vacuum-pack coffee jars; a quarter-million-jar order in 1932 | [S] #34 [@ev-0038, #34; @ev-0039, #34], Mackey collection [ev-0040], [H] [ev-0006] |
| Trappey’s Food Products / Tabasco Peppers | Louisiana | sauce bottles | [S] #70–71 [@ev-0038, #70-71; @ev-0039, #70-71] |
| Evangeline Pepper Sauce | St. Martinville, LA | pepper sauce | [S] #26 [@ev-0038, #26; @ev-0039, #26], [H] [ev-0006] |
| [[organizations/gebhardt-chili-powder-company | Gebhardt]] Eagle Chili Powder | San Antonio, TX | chili-powder bottles, 2 oz. and 6 oz. |
| Real Shine | — | shoe-polish container | [S] #57 [@ev-0038, #57; @ev-0039, #57] |
| Jay Bee Liquid Bluing | — | laundry-bluing bottle | [P] plate p. 51, embossed, absent from the price list [@ev-0038, p. 51] |
| Crown Brand | — | one-quart dill-pickle jar, surviving paper label | [P] plate p. 55 [@ev-0038, p. 55] |
The Crown Brand entry is weaker than the others in kind, not just in degree. Its evidence is a paper label on a stock Three Rivers quart, which documents a packer who filled Three Rivers jars; it does not document a jar made to that packer’s order [@ev-0038, p. 55].
The Gebhardt figure is the largest production number attached to any single customer and it comes from a newspaper report rather than from the company: about 61,000 bottles a day, enough to pack seven tons of chili in one day’s output [ev-0026].
Two of these rows are named in the licensor’s paperwork. Hartford’s tabulated register of its licences gives each licensee a permitted ware field, and the Three Rivers entry includes “proprietary and [m]iscellaneous ([she]llacs, shoe polishes and bluing)” [@ev-0059, part 18, line 67552]. Shoe polish and bluing are the two least self-explanatory classes in the table above, and both are here: Real Shine, a shoe-polish container [@ev-0038, #57], and Jay Bee Liquid Bluing, a laundry-bluing bottle photographed on a plate and absent from the price list [@ev-0038, p. 51]. Neither route knew of the other — a collector photographed two odd bottles; a licensor, years before, wrote the two ware classes into a schedule. The register names no customer and cannot add a row, since it does not say who bought bluing bottles. What it does is confirm independently that this plant was in those two lines of trade, which until now rested on two embossed specimens and nothing else.
And the two class names are Hartford’s, not the scan’s. A reader could reasonably ask whether “shoe polishes and bluing,” read out of a damaged column, is really what the register says. The same exhibit answers that from another licensee in another year. Its ledger of amendments records Pierce given, on 19 June 1936, “miscellaneous containers (ammonia, furniture polish, metal and shoe polish),” and six weeks later, on 31 July, “bluing, cements, cleaning fluids, dyes, fire extinguishers, sewing machine oil and turpentine” [@ev-0059, part 18, line 68242]. Both passages are clean in the scan. Shoe polish and bluing are Hartford’s own ware classes, written out in full, in the vocabulary a licence clerk used. That does not make the Three Rivers entry a clean read — its left margin is still shaved — but it removes the possibility that those two words were assembled out of noise, which is the failure mode a bad OCR reading of this kind would take.
Unconfirmed and negative cases
Suggestive but unconfirmed. The Hoffmann-Hayman “unknowns” section holds City Bottling Works with no city, a Coca Cola Bottling Works of Lockhart, Texas, a Smile of Glenmora, Louisiana, and an unlabelled clear 6½-oz. bottle with star marks [ev-0006]. None is counted above.
A customer named in a document rather than on a bottle. In January 1935 Frank C. Ball wrote to Hartford reporting Mason fruit jars being sold by the Griffin Grocery Co., which he believed Three Rivers had manufactured and sold to that firm. Hartford’s reply of 23 January recounts the charge and the receiver’s answer — that he had never made 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,” and that what a buyer did with caps afterwards was beyond his control [@ev-0059, Gov. Ex. 203, part 16, fol. 59030]. Griffin is not counted above. The claim is a competitor’s belief, relayed by the licensor, and answered; no specimen supports it, and the letter gives the firm no location. It is recorded here because it is the only customer name this expert holds that arrived on paper instead of on glass, and because it shows exactly what the embossed method misses: the disputed jars reached a grocery company, and no jar of that trade was ever going to be embossed “Griffin.”
A negative case worth recording. Texas Bottling Works of El Paso, active 1918–1934, used bottles from about 1924 that are identical to Union’s Three Rivers containers but carry no manufacturer’s mark. Lockhart does not count it a confirmed customer and neither does this expert [ev-0048]. It is the clearest illustration of the method’s limit: an unmarked bottle cannot be attributed, and Three Rivers did not mark everything it made.
Generic entries. Smith’s list includes beer bottles at #3–4 and plain milk bottles under #46 that name no customer at all [@ev-0038, #3-4, #46; @ev-0039, #3-4, #46]. They are the physical reminder that the embossed census is a floor.
What would extend this list
The roster is bounded by the method, not by the market. Every entry requires an embossed name and a surviving marked specimen and someone to have published it, and the sources that supply the third condition are concentrated in one distant city. El Paso is exhausted [ev-0058]; nothing comparable has been done for San Antonio, Corpus Christi, the Rio Grande Valley, or Houston, which are all far closer to the plant. A customer ledger, an order book, or a run of trade-press advertising would replace the whole method, and none is held.
The trial record shows a second route that would not replace the method but would run beside it. A competitor’s Dallas office was reporting Three Rivers jar prices to head office by size and closure [@ev-0059, Gov. Ex. 945, part 16, fol. 60283], and the licensor was corresponding with the receiver about who his buyers were [@ev-0059, Gov. Ex. 203, part 16, fol. 59030]. Fifty-four of the trial record’s fifty-eight parts have not been searched, and the exhibit series those two letters sit in is where a named jobber, packer, or grocery house is most likely to be found — customers this roster is structurally incapable of reaching from glass alone.
That route has now been searched as far as the holdings allow, and it is empty. The four parts held were swept for every occurrence of grocer, jobber, packer, canning, cannery, and the preserving and condiment trades, across the whole extract. The only named trade customer anywhere in it is the Griffin Grocery Co. recorded above; the exhibit series around Government Exhibits 945 to 949 and 203 names competitors, officers and prices, and not one buyer. So the lead is not disproved — it points at the fifty-four unsearched parts, which is where it always pointed — but it has been followed to the end of what this expert holds, and this roster gained no row from it. That is a search result and is recorded as one, so the next pass does not spend itself walking the same ground.