Industries / Food, farming and forestry / Food and beverage processing / Soft drink bottling
Syrups and mixing
Syrups and mixing is the work of making a drink's syrup from sugar or other sweeteners, flavourings and acids, then diluting it with treated water and dissolving carbon dioxide in it for the fizz, either at the plant or at a fountain where the drink is poured.
7
Milestones achieved, from 1886 to 2024.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1886
Syrup mixed at the fountain
On 8 May 1886 in Atlanta, the pharmacist John Stith Pemberton carried a jug of his new syrup to Jacobs' Pharmacy, where it was mixed with carbonated water at the soda fountain and sold for five cents a glass, according to the Coca-Cola Company's archive. Asa Griggs Candler, who took control of the business between 1889 and 1891, built it on making the syrup and selling it to fountains to be mixed with carbonated water there. The New Georgia Encyclopedia describes the company today as the world's largest maker and seller of soft drink concentrates and syrups, which others dilute and fill.
Credit
John Stith Pemberton; Asa Griggs Candler
1980s
Corn syrup replaces sugar
High-fructose corn syrup (HFCS) is a liquid sweetener made from corn starch, using an enzyme that turns part of the glucose into sweeter fructose; Richard O. Marshall and Earl R. Kooi first made it in 1957. After US sugar prices rose above 70 cents a pound in the late 1970s, Coca-Cola used HFCS in some minor brands and then phased it into its flagship cola in the 1980s, according to the trade publisher Sosland. In October 2025 Coca-Cola began a limited US roll-out of a cola made with US cane sugar in 12-ounce glass bottles, alongside the corn syrup version, and its chief financial officer, John Murphy, said only a certain amount of cane sugar is available in the United States.
Lessona drink's sweetener follows local crop prices and tariffs, so the same brand is made with cane sugar in Mexico and with corn syrup in the United States.
Credit
The Coca-Cola Company; Richard O. Marshall and Earl R. Kooi (the enzyme route)
1983
Aspartame cleared for sodas
Aspartame is a low-calorie sweetener that the body breaks down into two amino acids, aspartic acid and phenylalanine. The US Food and Drug Administration (FDA) approved it for carbonated drinks in July 1983. The path was not smooth: a public board of inquiry revoked the original 1974 approval in October 1980 over questions about brain tumours, and the FDA Commissioner overturned that ruling in 1981. In July 2023 the International Agency for Research on Cancer (IARC) classed aspartame as "possibly carcinogenic to humans", while the Joint FAO/WHO Expert Committee on Food Additives (JECFA) kept its acceptable daily intake of 40 mg per kg of body weight; the FDA said it disagreed with IARC.
Lessona sweetener's approval can be argued over again decades later, and two expert bodies can read the same studies differently.
Credit
G.D. Searle & Co. (developer); US FDA
2008
Stevia extract cleared
Rebaudioside A is one of the sweet compounds in the leaves of the stevia plant. On 17 December 2008 the FDA answered two notices, from the Whole Earth Sweetener Company (part of Merisant) and from Cargill, saying it had "no questions" about their conclusion that purified rebaudioside A is generally recognized as safe (GRAS) as a sweetener. A GRAS notice is the US route by which a company documents that an ingredient is safe and the FDA reviews it without a formal approval. It let US makers sweeten drinks with an extract from a plant.
Credit
Merisant and Cargill (notifiers); US FDA
2009
Fountains that dose cartridges
A fountain dispenser normally mixes syrup from 5-gallon bags with water and carbon dioxide at a fixed ratio. In July 2009 Coca-Cola began rolling out Freestyle, a touchscreen dispenser that offered over 100 drinks by dosing small amounts from concentrated flavour cartridges, each a little larger than a video cassette. It took four years and over $100 million to develop with DEKA Research and Development, Dean Kamen's company, and the machines were built to send their sales data to Coca-Cola every night. Coca-Cola's archivist says the precise dosing pump came from one of the company's earlier coffee projects.
Credit
The Coca-Cola Company; DEKA Research and Development
2018
A tax tiered by sugar
The UK Soft Drinks Industry Levy, announced in March 2016 and charged from April 2018, costs makers and importers 24p a litre for drinks with over 8 g of sugar per 100 mL, 18p for drinks with 5 to 8 g, and nothing below 5 g. Scarborough and colleagues tracked 209,637 supermarket listings and found that by February 2019 the share of affected drinks above 5 g had fallen by 33.8 percentage points against the earlier trend. Their data were not weighted by sales. Mexico's tax from 2014 instead charged 1 peso a litre on drinks with any added sugar; purchases of taxed drinks fell by 6% on average in its first year, by the estimate of Colchero and colleagues.
Lessona tax with steps by sugar content gave makers a reason to change recipes, not only prices.
Credit
HM Treasury (levy); Scarborough and colleagues (PLOS Medicine, 2020); Colchero and colleagues (BMJ, 2016)
2024
Brominated oil banned
Brominated vegetable oil (BVO) is vegetable oil bonded with bromine, used at up to 15 parts per million to keep citrus flavouring from floating to the top of a drink. The FDA took it off its list of ingredients generally recognized as safe in 1970 but let it stay "on an interim basis" while more studies were done. After FDA and National Institutes of Health studies found that the thyroid was a target in rats fed BVO, the FDA revoked the permission on 3 July 2024, effective 2 August 2024, with a year for companies to reformulate. By then, the FDA said, few US drinks still used it.
Lessonan interim permission lasted 54 years, so drink makers had long since swapped the ingredient before the rule caught up.
Credit
US FDA; National Institutes of Health
Watch for
A lower sugar line
The UK government says it will lower the levy's threshold from 5 g to 4.5 g of sugar per 100 mL and extend it to milk-based and milk-substitute drinks with added sugar, with makers given until 1 January 2028 to cut sugar. It says the levy has already cut sugar in affected drinks by almost half between 2015 and 2024, by its own figures.
Achieved whenthe lower line is in force and an independent study of shelf or sales data finds that drinks just above 4.5 g per 100 mL were reformulated below it.
Claimed by
UK government (HM Treasury and HMRC), November 2025
Sweet protein for drinks
Brazzein is a sweet-tasting protein first found in the fruit of a West African climbing plant. Sweegen says its brazzein, made by precision fermentation (microbes engineered to produce the protein in tanks), received an FDA "no questions" letter on its GRAS notice, announced on 25 June 2026, and that it pairs well with stevia to cut sugar while keeping the taste.
Achieved whena soft drink sweetened mainly with brazzein is on general sale, and an outside laboratory or tasting panel reports how its sweetness holds up over shelf life in an acidic, carbonated drink.
Claimed by
Sweegen, June 2026
Open
01
Fountain drinks against bottled
A survey method and open dataset that measure the sugar in fountain drinks poured at restaurants and cinemas, against the same brand's bottled or canned drink, and check diet pours for sugar that should not be there. Sugar is read as Brix, the percentage of dissolved solids by weight, with a refractometer, an instrument that reads it from how a drop bends light.
The briefCredit · unclaimed02
Sweetness of sweetener blends
An open model that predicts how sweet a blend of high-intensity sweeteners, such as sucralose, acesulfame-K and stevia extracts, tastes in an acidic, carbonated drink, given as the sugar concentration it matches, so a recipe developer can shortlist blends before a tasting panel.
The briefCredit · unclaimed
Open 01 · the brief
Fountain drinks against bottled
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A survey method and open dataset that measure the sugar in fountain drinks poured at restaurants and cinemas, against the same brand's bottled or canned drink, and check diet pours for sugar that should not be there. Sugar is read as Brix, the percentage of dissolved solids by weight, with a refractometer, an instrument that reads it from how a drop bends light.
Why it is still open
A fountain mixes syrup with water and gas at the outlet, so the result depends on how each machine is set and kept. Searches of Europe PMC and Crossref for this file found laboratory sugar measurements of bottled and canned drinks, such as Walker and colleagues' 2014 comparison of three laboratory methods, and studies of fountain drink prices, but no published survey of sugar in regular fountain drinks across many outlets. The nearest prior work is a 2009 conference abstract by Kessler and Kessler, "The Presence of Sugar in Fountain-Dispensed Diet Drinks", seen only as a bibliographic record. GitHub searches turned up no fountain or post-mix measurement project.
Built with what exists
A handheld digital refractometer with temperature compensation, a magnetic stirrer or ultrasonic bath to drive off the gas before reading, glucose test strips for the diet pours, a phone form such as KoboToolbox to log outlet, dispenser type and time of day, and Python for the analysis.
First step
Calibrate the refractometer against sugar solutions of known strength, then measure regular and diet pours at 10 outlets alongside bottled samples of the same brands.
Done looks like
Readings from at least 100 outlets in at least two cities, three pours per drink, for at least three brands, each compared with bottled samples of the same brand bought that week. The report gives the spread from the bottled value and the share of diet pours with measurable sugar, with all data public.
Open 02 · the brief
Sweetness of sweetener blends
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
An open model that predicts how sweet a blend of high-intensity sweeteners, such as sucralose, acesulfame-K and stevia extracts, tastes in an acidic, carbonated drink, given as the sugar concentration it matches, so a recipe developer can shortlist blends before a tasting panel.
Why it is still open
Drink makers blend sweeteners because some mixtures taste sweeter than the sum of their parts, an effect called synergy. A January 2026 Jiangnan University study built machine-learning models for blends, but only of sugars and polyols (sucrose, glucose, fructose, mannitol and sorbitol), and it says interactions in blends of several sweeteners remain "significantly underexplored". A 2024 study from a French public taste laboratory in Dijon and Cargill tested nine sweeteners and pairs of them on the human sweet taste receptor grown in cells, which is a screen, not a prediction of what a tasting panel would score in a drink. Wee, Tan and Forde (2018) published sweetness curves for 16 single sweeteners from a 40-person panel, and Schiffman and colleagues (2000) measured synergy in three-way mixtures of 14 sweeteners, both in plain water. No open model for high-intensity blends in soft-drink conditions turned up in literature or GitHub searches.
pmc.ncbi.nlm.nih.govpmc.ncbi.nlm.nih.govpmc.ncbi.nlm.nih.govebi.ac.uk
Built with what exists
The published curves and mixture results as a starting set; a trained panel of 12 to 20 tasters selected under ISO 8586; food-grade sweeteners at permitted levels in a lemon or cola base with citric or phosphoric acid; scikit-learn or a small Gaussian process model, with active learning to pick which blends to taste next.
First step
Fit a simple additive model to the 2018 curves, then run a 12-person panel on 10 two-way blends of sucralose and acesulfame-K in carbonated lemon water and measure how far the additive guess misses. Tasting panels at a university usually need ethics approval.
Done looks like
On 30 blends not used in training, the predicted sugar match falls within 1 percentage point of sucrose of the panel's value for at least 25 of them. The panel data, recipes and code are public.