Home | Membership | News | Diary | Walks | Calvocoressi Fund | Books | Journals | Links | Database | e-papers | About us

Acetone production at Nicholson's Distillery, Three Mills, Bow

Martin Adams, Patrick Graham and Brian Strong


The story of Chaim Weizmann's work to develop a fermentation process to produce acetone from grain and the use of Nicholson's Three Mills Distillery as a pilot plant during the First World War is relatively well known and referred to in a variety of sources. 1 This paper is based on documents from Three Mills which have only recently come to light, coupled with research at the National Archives and is intended to recount some of the rather more complex story behind this and expand on the role that the Three Mills Distillery played (Figure 1).

During the summer of 2010 Catherine and Jon Battell donated a number of folders of documents to the River Lea Tidal Mill Trust (RLTMT). The Trust is the owner of The House Mill, a large tide mill on the lower River Lea at Bromley-by-Bow, London E3. The mill ground grain for the adjacent Three Mills Distillery, a major producer of potable alcohol for the gin trade. The distillery has been converted to other uses now and the donors considered the Trust, with its interest in the history of the area, a suitable recipient of the papers (Figure 1).

The folders appear to have belonged to Catherine's grandfather, Richard Ernest Blundell, who was a brewer/distiller at Three Mills from 1923 until the bombing of Three Mills in October 1940. Two of the folders relate to the work carried out there in World War I on the production of acetone from grain.


Acetone is an organic solvent that had an important role in the production of cordite, a smokeless propellant developed in Britain in 1889 for use by the military. It was obtained by the destructive distillation of wood and the bulk of the UK's supplies came as imports from the United States, Canada and Austria. The hugely increased demand for cordite from the start of the First World War and the loss of one supplier (Austria) to the other side of the conflict led to a serious shortfall in supplies of acetone.

To address this shortage, attention turned to alternative sources of acetone, in particular its production by a bacterial fermentation process. This had been identified as a possible source some years previously by a group of scientists and technologists based around a London firm of technical research chemists, Strange and Graham Ltd, and a research group based at Manchester University under the leadership of W H Perkin. However, production from a factory in King's Lynn using their process proved disappointing and it was at this stage that Chaim Weizmann was approached following his earlier offer to the government of a similar but allegedly superior process. Weizmann had previously been a key member of the Manchester research group under Perkin but had been dismissed following a dispute in 1912 (Figure 2). He continued working independently on the same line of research, in the process isolating a bacterium, Bacillus BY, now known as Clostridium acetobutylicum, which proved far superior to the organism being used by his rivals. One important advantage was its ability to utilise maize as opposed to potatoes as its raw material.

In April 1915 Weizmann received a summons from Sir Frederick Nathan, then Admiralty advisor on cordite supply. 2 During this visit he met the then First Lord of the Admiralty, Winston Churchill, who famously asked him, 'Well Dr Weizmann, we need thirty thousand tons of acetone. Can you make it ?' 3 Weizmann, by his own admission unnerved by such forthright questioning, was at first hesitant saying that he worked in a laboratory and so far had only succeeded in making a few hundred cubic centimetres of acetone at a time by the fermentation route. But, presumably regaining his composure, he went on: 'if I were somehow able to produce a ton of acetone, I would be able to multiply that by any factor you choose. Once the bacteriology of the process is established, it is only a question of brewing.' 4

The Clock Mill and Three Mills Distillery

The obvious next step was to establish a pilot plant to test the process on a larger scale and this is when the Nicholson's distillery at Three Mills enters the story. On that same day (21st April) Weizmann visited Three Mills.

Work at Three Mills

Between July and September 1915, starting in a small shed equipped with a 2000 gallon fermentation tank and a small still, Weizmann conducted a series of trials which showed that it was possible to produce fermentation acetone on a reasonably large scale. Weizmann in his autobiography notes that they did run into lots of trouble along the way when trying to simplify and translate the laboratory process into something more suitable for industrial production. Among the shortcuts they tried was an attempt to forgo any kind of asepsis (measures to exclude microbial contamination) from the process as these were costly and time consuming. This was, however, a short cut doomed to failure. In a report dated 13th October 1915 Weizmann reports:

Crucially, he also noted that his third point was extremely difficult to fulfil under the existing conditions at Bromley. 5 While the factory had successfully operated as an alcohol distillery for many years, the traditional yeast-based fermentation employed was quite robust when it came to resisting contamination problems. This was not the case with the much more sensitive and demanding Bacillus BY used in the production of acetone.

Manchester University Chemistry Department 1911. Weizmann is seated front row fifth from the left. W.H. Perkin is seated immediately to his left. (Copyright of the University of Manchester)

By this stage however the success of the Three Mills pilot trials had led to the decision to expand production by fermentation to meet the deficit in acetone supplies. The Admiralty started to build a purpose-built plant at Holton Heath, Dorset, near their cordite factory, and David Lloyd George, the Minister of Munitions, had announced at a meeting with alcohol distillers in February 1916 that distilleries would become controlled premises and that, after accommodating the need to produce yeast and industrial ethanol, the whole of their capacity would be required to produce acetone. 6 On 14th March 1916, the King's Lynn factory was also taken over to become HM Factory King's Lynn and production switched to use of the Weizmann process.

Information from the Three Mills files

The material in the two folders given to the River Lea Tidal Mill Trust relates to the work on acetone production at Three Mills in the period from August 1916 to March 1917. By this time the process had been expanded to other sites and Weizmann was now working for the Admiralty at laboratories in the Lister Institute. A letter to Weizmann in the National Archives dated 5 April 1916 describes how his duties were now to focus entirely on his work at the Lister Institute where he was to train chemists in the bacteriological skill necessary for them to work in the distilleries and to produce cultures to be sent out to distilleries around the country. 7 His name occurs only twice in the Three Mills folders; once in reference to a 'Weizmann pail', and he did not even attend an important meeting at Three Mills on 6 November 1916 to discuss the problems of contamination which were occurring in the process.

One folder labelled 'Rushton's Weekly Report' appears to be copies of the reports sent weekly to the Ministry of Munitions from the week beginning 10 September 1916 to 13 January 1917 by Mr W Rushton. He was an employee of the Ministry of Munitions based full-time at Three Mills and was designated 'chemist in charge'. The more instructive folder is entitled 'Acetone, Three Mills' and bears the name J W D Drake (a brewer/distiller at Three Mills along with Ernest Blundell, Mrs Battell's grandfather). Among its contents it describes the equipment used, the production process, the labour force involved (35 different names are listed, all apparently male) and some of their duties. It also gives a vivid impression of the the many problems which arose during operation, particularly those associated with infection by contaminating microorganisms.

Acetone production process using Lister Institute culture

From the information in the folders it is possible to reconstruct the process as applied at Three Mills and this is illustrated in Figure 3. But it is also apparent from the files that this was still developmental, subject to change, and quite a long way from being reliable and reproducible. Numerous planned and enforced modifications were continually being made. A rather confusing mixture of equipment: purpose built, adapted and existing plant from the distillery was being used and the progress of a fermentation run was rarely routine. It was a new and difficult process and experience from traditional fermentation industries such as brewing and distilling was of only limited help.

The process at Three Mills

As shown in Figure 3, the process has one hallmark of a modern industrial microbiology process in the sense that the organism (Bacillus BY) is grown up through a series of stages starting from a primary culture (generally supplied by Weizmann's laboratory at the Lister Institute). This is the sequence that runs along the bottom of the diagram from left to right. At each stage, increasingly large volumes of a sterile medium are inoculated with an active culture so that substantial inoculant or seed of active bacteria is available to start the next, still larger, stage. A large starting population of bacteria helps ensure a rapid and successful fermentation. The volume of inoculant used is at least 1% but varies between 1% and 13.3% for different stages. The sequence ends ultimately with inoculation of what are called the washbacks, 30-40,000 gallon tanks, the product from which is then distilled to separate the acetone.

Acetone production process using King's Lynn culture

What is apparent at first glance from Figure 3 is the multiplicity of different vessels in use: seed pails of different sizes, larger seed tanks, Bub vessels, two tanks known as the Zeppelin (the first Zeppelin raid on London had taken place in May 1915) and the Hindenburg (Chief of the German General Staff from 1916), and several washbacks. The washbacks were simple iron vats of 30,000 or 40,000 gallon capacity with wooden lids and a six inch pipe for gas discharge and to prevent formation of a vacuum after steaming. They could not stand steaming under pressure. There was a small opening in the top (closed by a lid) used for charging the vat and for sampling and temperature monitoring. Wood as a construction material is anathema to the modern microbiologist due to the impossibility of cleaning and sanitising it effectively, but this is recognised in the papers and it is noted that iron lids, pressure gauges and safety valves were being fitted to two new washbacks capable of withstanding a pressure of 1-2 lbs.

The medium in which the organism was grown during preparation of the seed cultures was a mash of 5% ground maize in water. This was sterilised by pressure cooking at 20-30 lbs for between 2 and 3 hours. In the final stages of the process, the medium used in the Hindenburg and in the washbacks in which the bulk of the acetone was produced was substantially different. Described as wort (a brewing term for a fermentable sugar solution), its preparation was similar to the production of the wort used in brewing and distilling and would have been more familiar to those working at Three Mills. The inclusion of malt meant that the carbohydrate was present as maltose and dextrins rather than the polymeric starch present in the maize mash in which the inoculum had been grown. In an extended brewing report by Mr Drake (16 October 1916) Weizmann gets his second mention in the files when Mr Drake notes that 'Weizmann thinks that both maltose and dextrin are fermentable by the B.Y. organism'. Why a different medium was used in the final production stage is not clear.

In preparing the wort, malted barley was added to the mixture of maize along with some oats in hot water. The mash of grains and hot water was prepared over a period of 3 hours, followed by 2 hours standing to allow the malt to break down the maize starch into maltose and dextrins. The wort, a relatively clear liquid, was drained off from the grain residues which were then washed with hot water a further 3 times to remove all the residual sugars. The resulting wort was boiled for two hours before cooling and inoculation with the culture. In the Hindenburg, for some unexplained reason, the wort was supplemented with oats (5 cwts to 10,000 gallons) as a feed supplement and then boiled for 2 hours at a pressure of 1 lb (cooling took 4-5 hours). The residual grain solids in the mash tun were sold for animal feed at 35/- (£1.75) per ton.

Apart from the initial test tube cultures which were incubated for 2 days, the other stages up to the inoculation of the washbacks were incubated for 1 day each, so that it took 5 or 6 days between first inoculation of the tubes and the seeding of the washbacks. Once the washbacks were inoculated the fermentation took about 48 hours. Samples were taken periodically by means of a dipper let down from an opening in the top of the washback and the progress of the fermentation monitored by measuring the acid produced during its first phase and the rate of gas evolution. The microbiology was monitored principally by microscopy — simply looking at the organisms present down a microscope, although some plating out of cultures was also done. When the fermentation had finished, the wash was pumped into the wash chargers and thence into the still where the acetone was separated from the butyl alcohol which was also produced in the process.

The problems at Three Mills

In the files, there are several technical reports in which comparisons are made between the process as applied at Three Mills and those at the King's Lynn factory and at the Ardgowan distillery in Scotland where acetone was also being produced. From these it is clear that Three Mills may have been the first plant to produce acetone by fermentation using the Weizmann process but it was not the most successful or reliable. Difficulties in preventing contamination of the process with unwanted microorganisms was the single most important factor in this. The problem was apparent to Weizmann in his 1915 report cited above and is reiterated in comments in the files almost exactly a year later (September 1916) by a Mr Appleyard from the King's Lynn factory. He makes the point that Bacillus BY cannot withstand competition for food. The principal source of these competing contaminants is the maize itself (he mentions two principal organisms Bacillus Butyricus and Bacillus Guntheri) so, he continues, it is essential that all the media and vessels must be completely sterilised prior to use. Boiling wort for 2 hours would not necessarily do this. 'Our experience at King's Lynn and other places has convinced us that, to carry out the process successfully, a sterile medium and sterile vessels are absolutely essential. If sterile conditions are absent the mash is worthless for the production of Acetone.' Frequent mention is also made in the Three Mills papers of a serious contaminant known as Streptococcus X. It is not possible to say what this microorganism would be described as today, but it is likely to have been what is known as a lactic acid bacterium that would produce large amounts of lactic acid and effectively poison the fermentation. This is not as heat resistant as the Bacillus species mentioned above and lends support to the notion that inadequate sterilisation was a common failing.

Fermentations were also tried at Three Mills to compare the efficacy of BY cultures from King's Lynn with those from the Lister Institute. A Dr Hose, superintendent of the King's Lynn factory, attended a number of meetings at Three Mills where his role seems to have been that of an expert advisor. Other experiments were conducted at Bromley at the suggestion of Dr Hutchinson from the Ardgowan distillery.

It is apparent from the reports that total or partial failure of the fermentation was a fairly common occurrence. Varied and numerous incidents are recorded such as a steam valve being left open by accident killing most of the BY bacteria during an active fermentation or a mixture of formalin and carbolic acid being sucked into one of the fermentation vessels with a similar, but less severe, effect. There is also a litany of minor hiccups causing potentially damaging delays in inoculating vessels such as bursting steam pipes, a rouser belt breaking and misunderstandings between staff.

Most problems arose from failure to control contamination from microorganisms in the environment, the raw materials and inadequately sterilised medium or equipment. One example of this is the instance of non-sterile air being sucked into vessels as they cooled. In another, one of the weekly reports mentions that during preparation and fermentation of the Bubs, the floor of the Bub house was being dug up for concreting and the earth being dug up was badly infected and smelt very bad. Another records that in sterilising one Bub vessel, the packing of the lid was found to be rotten.

Under these conditions the need, at each stage, to inoculate several different vessels to produce sufficient seed for the next stage increased the risk of contamination substantially as did hold-ups and delays caused by problems earlier in the process. From Figure 3 we can see that at one stage seven different vessels are used simultaneously to produce seed culture while in the following step three are used. An alternative process flow tried at Three Mills and described in the papers used a culture from King's Lynn (Figure 4) and is also significantly less complex with a single vessel being used at each stage of seed preparation. This may have been replicating a simpler process as applied at King's Lynn, and the point is made in a letter from Dr Hose to the Propellants Branch of the Ministry of Munitions that the number of cultivation steps used at Three Mills should be reduced to a minimum.

The accompanying report by Hose and Appleyard, under the heading: 'The condition of the plant at Bromley with regard to sterilisation', states:

Despite these shortcomings, Three Mills was still producing appreciable quantities of acetone throughout this period. It was sent principally to the Royal Cordite Factory at Holton Heath, Dorset but on one occasion in the period covered a small consignment was sent to the Royal Gunpowder Factory, Waltham Abbey. Certification letters signed by Mr Rushton the chemist in charge confirm production of 2977 gallons (nearly 11 tons) of crude acetone (2269 gallons pure) in the period 31 May to 30 November 30 1916, 6103 gallons (4966 gallons pure) between 30 November 1916 and 31 January 1917, and 7267 (5781 pure) in the month 1-28 February 1917.


Despite the rapid development of the acetone fermentation into an efficient industrial process other exigencies of war intervened to halt its expansion in the UK. The maize used had to be imported taking up valuable shipping space at a time of submarine warfare and blockade. Imaginative attempts to find alternatives to maize led in the summer/autumn of 1917 to a nationwide campaign to collect horse chestnuts for this purpose. Plans were also laid to transfer production to maize producing regions. The Canadian distillers Gooderham & Worts were recruited into this effort in 1916, and were producing 1000 tons of acetone per year by the war's end. When the United States entered the war distilleries were also acquired in the US and the basis for a substantial successful post war industry was formed. Outside of wartime, focus shifted to the other major product of the fermentation, butanol, which was required in the production of lacquers used in the burgeoning car industry. With the advent of cheap oil and the growth of the petrochemical industry in the 1950s, the process fell from favour in most countries although recent years have seen some resurgence of interest, particularly in China.


We are grateful to Catherine and Jon Battell for donation of the documents that contributed to this article and to Professor Mike Bushell for his valued assistance with the process flow diagrams.

The authors

Patrick Graham is a retired chemical engineer and long time member of GLIAS. He was a volunteer at the House Mill for many years and has previously written for LIA on another local site, Kemball Bishop.

Brian Strong is a retired civil servant. His particular interests are the London Distilling Industry and Tide Mills. He is a former Trustee of the River Lea Tidal Mill Trust, which owns the House Mill, and is a volunteer for House Mill openings. He has been secretary of GLIAS since 2001.

Martin Adams is Emeritus Professor of Food Microbiology at the University of Surrey and Editor of the GLIAS journal, London's Industrial Archaeology.

Notes and references

1. Jim Lewis, Industry and Innovation: the technological revolution in the Lea Valley 2010: Libri Publishing, Faringdon, pp 85-88

2. Lewis ibid. and Weizmann's autobiography both have these events occurring in 1916. Dates on documents in the National Archives clearly indicate that work was already in progress in 1915

3. Chaim Weizmann, Trial and Error 1949: Hamish Hamilton, London, p 220

4. Weitzmann ibid.

5. Weizmann's Reports National Archive MUN 7/236

6. Minutes of conference 11 Feb 1916 National Archive MUN 7/240

7. Letter to Weizmann dated 5.4.16 MUN 7/239

© GLIAS, 2015