We live, We learn … About Lactics

The herd supplying Neal’s Yard Creamery (also a herd with some Montbeliard influence) lining up to be milked.

As prospective cheesemakers working with a milk supplier usually do, Rose and I have been sending off milk samples for microbiological testing for some months now.  While we’ve had generally good results regarding absence of pathogens, I was taken aback to discover that we also seemed to have an absence of lactic acid bacteria…or certainly we had a lot less then we wanted.

‘Most milk in the UK now is not good for cheese,’ pronounced Ivan Larcher at one of my SAF courses, ‘It is dead milk.’

‘A little damning, surely,’ I thought.

Lactic, if my dictionary is to be believed, means ‘relating to or obtained from milk’. It derives from the latin ‘lactis’ genitive form of the word ‘lac’ or milk.

Doesn’t milk just HAVE lactic acid bacteria in it?

Well, apparently not always and if it does, 1 day in a bulk tank and they are not very happy bugs.

One side of the milking parlour at Merrimoles Farm showing the stalls in which the cows stand, the pipework and the clusters below.
For any of my non cheese industry friends reading this post. First off, well done for making it this far into a pretty cheese-geek technical post and good luck with the rest of it. This, should you be wondering, is a cluster and it goes on the cows udders.

Our milking system is like most in the UK Dairy industry.  It has a series of clusters along parallel milk pipes.  Vacuum pumps pulsate to remove the milk from the cows udders and it is piped out of the parlour, through a filter, then a plate cooler and finally into the bulk tank where it’s held at 4C until the lorry comes from Dairycrest to collect it.   They come every 2 days and collect 4 milkings.

The Bulk Tank.

When we test, we take our samples from the outflow pipe of the bulk tank.  We also send off the milk filter from the last milking which the lab immerse in water and then test the rinsate.  We don’t know anyone else testing the milk filter so when we got our first set of results back and discovered literally millions of bacteria on it, we didn’t know whether this was normal, really bad or even really good.  We certainly knew that millions of the little critters looked pretty alarming on paper.

Our bulk tank milk samples seemed to show a happy grown of Pseudomonas (we don’t want happy Pseudomonas) and a rather less happy growth of Lactic acid bacteria.  The milk filter results seconded this.  In the autumn, we called Ivan back for advice and subsequently did a big clean through using peracetic acid.  Following that we have used a weaker peracetic acid solution for the final rinse of our pipework.

To start off 2014 in the way we intended to continue, with some more testing.  This time, we were hopeful for better results and to be a bit flirty we were going to get Andrew the kindly milker to hand milk a couple of cows for us to see how they compared for lactic acid bacteria.  These cows were a black and white cow that looks more Friesian Holstein in appearance who goes by the name of 266 and a brown and white one that looks more Montbeliard in appearance with the name 258.  266 was docile and calm when milked.  258 was disconcerted not to be in her usual clusters and stamped about a bit. We could identify the test results later because Andrew got less milk from her before asking if he could give up.  The cows that were hand milked were only given a dry wipe to their teats before milking.  Normally when they are milked into the parlour, their teats are given a wash before the pumping starts.  It has reduced the total bacterial counts of the milk right down but rather unfortunately we think it may be washing off our lactic acid bacteria.

The bulk tank contained 3 milkings at this stage.  The fourth was about to take place as we tested and Dairycrest were due that night to collect.   The milk filter was from that morning’s milking.

The samples were posted, results duly came back and we emailed Paul Thomas for advice and guidance in their interpretation.



The bulk tank results were better than we’ve had at other points in the past in terms of Pseudomonas.  Before now we have had counts of 21,000 per ml of milk.  This time we had a count of 170 per ml.  However it did show us some staph aureus too which is less than ideal.  According to our lab (Microtech Wessex) we would hope to see around 80% of the total bacterial count being lactic acid bacteria and unfortunately still on this sample it is considerably less than that.  With a total count of 53,000 total bacteria per ml of milk, this would mean we’d like to see 42,400 of these to be lactic acid bacteria.  According to the test result there are 1,200.

The hand milked samples were very different one from the other.  No 266, the black & white cow’s sample was extremely low in everything.  Almost nothing grew on the lab plates according to the lab.  Its counts are 0 in everything except yeasts.  So it has no coliforms, pseudomonas or staph aureus but unfortunately no lactic acid bacteria either.  I imagine Dairycrest would love it.  For making cheese it isn’t ideal.

No 258, the brown and white cow’s counts however were about spot on what we want.  It had a total count of 8,800 total bacteria which isn’t huge.  However, according to our ideal 80% we would be hoping to see around 7,100 lactic acid bacteria and we have 7,040.  There are 10 yeasts which is good, no Pseudomonas, no Staph aureus.

The two cows results, while interesting and raising a few interesting ideas, thoughts and questions only really give us a snapshot of the milk of 2 animals.  Importantly, though, it does show us that we have got the right balance of lactic to everything else in some of the animals in our herd.  Interestingly, Paul suggested that according to some of the papers he has read, the animals that line up to be milked first, being in general the livelier and healthier animals of the herd, often have lower somatic cell counts (an indicator of health) and as a result often have higher amounts of lactic acid bacteria.  If, as we are planning to do, we take our milk from the animals that are milked first, not only will we be using the pipelines at their cleanest but we will also be getting milk that is better suited to our cheese.  In addition, we will be taking the milk away without it being cooled.  This makes sense from an energy standpoint – why cool it to heat it back up to 38C – but also allows the lactic acid bacteria to compete with the Pseudomonas.  Cooling the milk to 4C and then storing it at that temperature for 36 hours has stopped the small numbers of lactic acid bacteria growing and reproducing but at those temperatures, the Pseudomonas can still grow.  According to survival models Paul referenced we could probably knock that 1200 per ml down to 120 just due to the storage time at cold temperatures.

Milk Filter:

The Milk filter results have always looked rather alarming to us in terms of Pseudomonas.  But corresponding with Paul Thomas meant he helped us by analysing the results so that we can compare them more easily against our milk results.

The relevant results are (per filter):

  • 1,500,000 Pseudomonas
  • 10,000,000 Enterobacteriaciae

So – big numbers.  But, as Paul said, we have to interpret them based on the amount of milk that has gone through that filter.  The most recent milk report I have from the farm dates back to December but if levels are similar to those in December’s monthly report, we are looking at 120,530 litres of milk for the month.  This means a daily total of 3,888 litres.  Each milking there’s a new milk filter so while I expect there’s a difference in quantity between morning and evening, for the sake of mathematical ease, let’s say half of that quantity is applicable to our filter tests: 1944 litres.  1944 litres works out at 1,944,000 ml which has gone through the filter.

Assuming the filter removed 50% of our bacteria, this then suggests that before filtering the total quantity of milk (all 1,944,000ml of it), it contained:

  • 3,000,000 Pseudomonas.
  • 20,000,000 Enterobacteriaciae

So per ml of milk we have

  • 1.5 Pseudomonas and
  • 10 Enterobacteriaciae.

Which makes it all look rather a lot better.

Even if we assume the filter only removed 10% of our bacteria, this still suggests that the levels in the milk weren’t huge.  If that were the case, we’d be assuming pre filtration numbers of:

  • 15,000,000 Pseudomonas
  • 100,000,000 Enterobacteriaciae

So per ml of milk we have

  • 7.7 Pseudomonas
  • 51.2 Enterobacteriaciae

However this doesn’t take into account the fact that the filter had been in the parlour during the day and we didn’t have fully frozen ice packs in our insulated box.  The sample was 10C when it was tested and apparently we can knock at least a couple of zeros off our total counts on the filter based on the time it had rested at that temperature since milking and whilst being posted to the lab. All of a sudden, this makes our Pseudomonas and Enteros presence not alarming at all.



The reason the bulk milk samples have been high in Pseudomonas in the past is that they are able to grown at 4C whereas lactic acid bacteria aren’t.  If there aren’t that many lactic acid bacteria in the milk in the first place and even on our brown and white cow friend 258, there weren’t huge numbers then the bulk tank is the worst conditions for them to grow and the best conditions for something that is happier at cold temperatures to get a head start.  Lactic acid bacteria like a range of temperature around body temperature basically but can grow from 20C – 50C.  So the bulk tank is giving an advantage to the wrong bacteria for cheesemaking.  In other words it is entirely worth it to arrange for the pipework we are planning and have our cheesemaking milk taken off before it goes through the plate cooler, and not just because of the energy use considerations.

There is still the cluster wash and some of the pipework which remains a concern as it will reduce lactic acid bacteria and potentially if there is doubt about the cleanliness of the water, will add in pseudomonas and possibly listeria.  In order to investigate this, we need to do a further milk test or water test.  I don’t know if it’s at all possible for the cluster wash to be switched off ever?  I am imagining not but it’s worth asking.  For the cheesemaking, it’s all about balancing the lactic acid bacteria against the pathogens and spoilage bacteria and the better we preserve the lactic acid ones the less we worry about the others.

Staph aureus:

The test results on this occasion aren’t hugely helpful but they basically indicate that they are present in the bulk tank and passing through the milk filter.  Evidently they aren’t present on every cow as there were none on either hand milked cow.

Paul Thomas’s theory based on some studies he has read (but I’d have to ask him if you wanted to know which ones) is that the animals that line up to be milked first tend to be the more vigorous, healthier ones which will be less likely to have Staph. aureus infections even at a subclinical level.  The less healthy ones will lag behind.

This suggests that our idea of taking the first bit of milk that goes through the milkline is probably a good one from the point of view of getting milk that is better suited to our cheesemaking.  Interestingly the animals with lower somatic cell counts (according to Paul), also tend to be the ones with higher counts of lactic acid bacteria as well.  Again, I’d have to push him for which papers supported that theory but it seems to indicate again that we will get more suitable milk for the cheese if we take the first lot of milk rather than from later in the milking.  Which means I will be getting up bright and early to collect.

A possible thing we could investigate as well is to look into the mastitis records to see if there are any patterns.  Paul (again) has had previous experience where with his milk supplier’s animals each cow that developed mastitis got it on the same quarter for a whole 2 week period.  It turned out that there was a contaminated rubber on one of the clusters.

Lactic Acid Bacteria:

The hand milk results do show that on individual animals we have pretty much perfect milk provided we can then manage the process so that we can get hold of that.

There’s no scientific basis for this that I know of but it’s a commonly held opinion that Friesian Holsteins are not as good producers of lactic acid bacteria and other breeds like Montbeliard are better.  Interestingly our results showed the perfect milk from a brown and white cow that Andrew felt would have more influence of Montbeliard in her genetics.  I don’t know how true that is however and it’s something to try and find out more about.

By taking milk from the livelier, healthier first milkers and keeping that milk warm we’ll give the growth advantage to the lactic acid bacteria as well as the other organisms.  Conventional wisdom sounds a loud klaxon at this point and shouts

‘What about growth of pathogens??? Re-frigerate!! Re-frigerate!’

And were it in isolation with no lactic acid bacteria, they’d be right to be cautious.  But by keeping the milk warm we are giving our lactic acid bacteria an even chance to consume that lactose and reproduce.  Then, when we add our starter cultures into the mix as well it should mean that lactic acid bacteria as a proportion of the total bacteria as we start to make cheese, out competes any pathogens or spoilage organisms.

The hand milking results also reassure us that the milk when it hasn’t gone through the cluster wash system does have enough lactic acid bacteria in it to try the experiments of making our own starter cultures from the milk.  This has always been an aim of ours which we thought we would have to postpone for a year or so at best but now seems much more possible.  It is, however, a discussion for another day and will involve a lot of hand milking and some careful selecting of suitable cows.

So, at the end of a rather head-hurting few days of analysing, emailing and thinking very hard, it’s good news.  We can try out making starter – hooray!! Obviously we won’t be using it unless it passes micro testing but for a while it didn’t look like we’d be able to even try.

It’s also a new list of questions to research.  Do the brown and while more Montbeliard looking cows like 258 give us better milk for our cheese?  Do the black and white cows have less lactic acid bacteria?  Does it all relate to their Somatic cell count levels?  Does the cluster wash still remove too many lactic acid bacteria and can there be an alternative?  If we can hand milk cows to make our own starter, does that matter?  And so on and so on.  Then there’s an off the wall ideas that Paul suggested too.  There are some studies in humans indicating that before giving birth, the nipple duct microflora is influenced by apparently deliberate movement of bacteria from the gut to the mammary gland by dendritic cells.  Perhaps this happens in all mammals and may account for the transfer of lactic acid bacteria into the baby’s and calf’s stomach with colostrum?

Who knew milk could be this complicated and this fascinating?  Just as well I never wanted a quiet life.

The Lightbulb Moment

One of the exciting things about working with Merrimoles Farm on the Nettlebed Estate is the potential inherent in their milk.  The herd is a mixture of Friesian Holsteins with Swedish Red and Montbeliard bred into the herd for increased vigour.  Increased vigour is the immediate benefit for the farmers along with better health, less likelihood of lameness and better fertility.  However it isn’t just a benefit to the farmer.   The cheesemaker (that would be me) naturally gets better milk from healthy, happy animals but the breeding with Swedish Red and Montbeliard is exciting because the solids in the milk of both of those breeds lend themselves more to cheesemaking than that of the Friesian Holstein or Holstein itself.  In addition to that, the animals graze on organic pasture and, as a result, spend most of the year happily outside, munching grass, flowers and herbs in the fields, which in theory means that they should have a diverse diet which will lend to aromatic compounds in the milk and the potential for a diverse grouping of lactic acid bacteria derived from the bacteria present on the teats.
I had come into the discussions with Rose and with Merrimoles Farm with a mental ticklist of what I was looking for in a milk supplier, namely, interesting breeds (check), outdoor grazing (check), varied pasture (check), preferably organic or as near to as possible (check).  Part of this is ideological, I don’t want to be involved in an enterprise where there are unhappy animals and where the farming isn’t sustainable and respectful to the environment.  Part is flavour driven.  Interesting breeds, varied pasture and organic management of the herd and pasture should, in theory, translate to the most interesting milk.  In other words, if I can unlock it, there’s a lot of potential for good flavours in our cheese.
‘Our milk is really good’ Rose told me proudly when we first met.
We collected a sample from the bulk tank to drink it fresh and (naturally) unpasteurised.  It tasted lovely.  I agreed with her.  On other occasions that I have drunk it since then, it consistently tastes lovely with a milky sweetness, mineral undertones and a velvety creamy mouthfeel.
‘And we get really great test results too’ she continued, ‘Dairycrest actually say that our counts are really low.  For an organic farm it’s practically unheard of.’
When I first visited the farm, Phil, the farm manager showed me a printout of their milk results which included fats and protein content as well as their routine total bacteria counts.  Having looked over other milk results in the past when I was Quality Assurance Manager at Neal’s Yard Dairy, I too was surprised at how low some of the counts were.  I expressed my surprise to Phil also who confirmed that yes, they were often told how rare it was for an organic farm to hit those levels.  It was something they were all proud of and justifiably so.  For their current customers this is exactly what they need and want.
I was a little more cautious.  Low total counts, seemed to me a good starting point, but more important than that is what that total count breaks down into.  Ideally, of course, that total count is entirely composed of lactic acid bacteria.  Worst case scenario, it’s entirely composed of Listeria monocytogenes or another pathogen.  As we drove away from the farm, I mentioned to Rose that, before we got into cheesemaking, we needed to build up a history of testing the milk in more detail.
‘For raw milk cheese,’ I explained, careful not to cast aspersions on what was evidently, very carefully produced milk, ‘It’s not so much the total counts we’re concerned about, but what’s in there.  So we need to send off some samples for testing and cover all the pathogens: Listeria monocytogenes, Staph. aureus, E.coli O157 and Salmonellae.’
It wasn’t urgent to get started testing straight away and actually we are hardly going to go and find another milk supplier.  Our cheese business is being started to make better use of the Merrimoles milk.  Whatever the results, we were already committed to working with them, so to begin testing nearer our production time made more sense.  I took a bottle of milk away to do a lactofermentation at home.  They took 48 hours to set, tasted yoghurt-like, although a little bitter, and had about one gas bubble.  I would have been happier to find no gas bubbles and to not have tasted the slight bitterness, but I wasn’t too bothered at this stage.
When we did begin our testing in May this year, however, we had a bit of a shock.  The total counts were low, as usual, but the lactic acid bacteria counts as a proportion of that were also a lot lower than we had hoped for.  We thought, especially as we want to culture our own starters, that we would be aiming for 80% of the total count to be lactic bacteria.  We were finding considerably less than that.
This means for our cheesemaking, culturing our own starters is a project for a few years time and won’t be happening initially.  If you make your own starters, the argument goes, you will have a more diverse population of bacteria but, of course, you don’t know what you are getting and they are likely not to acidify as strongly as bought and proven starters.  Just at the moment, we need to use plenty of proven starter to get our milk to acidify.  This is fine, I can work with that and still do my best to use a varied and interesting cocktail of cultures.  What makes me a little nervous though is that not having a naturally strong population of lactic acid bacteria does mean that we don’t have a built in safety mechanism in form of the milk’s natural ability to out compete pathogens.  If something nasty gets in, it can have a little pathogen party, reproducing itself all over the shop.
‘What does this mean for your cheese?’, I hear you ask.
Well as I said, we’ll use bought in starters and in addition we will record our acidity curves with every make.  We will also prepare ourselves for higher testing costs as we will have to test each batch that doesn’t acidify quickly enough and higher wastage for the cheeses whose test results don’t make the grade.  In the longer term, we’ll begin learning a lot about the factors that encourage or discourage lactic acid bacteria, because, in theory, with organic production and grazing outdoors, we should have plenty of them and yet we don’t.  With that in mind, at Slow Food’s Cheese this year, I listened avidly to their workshops on milk production.  But that is a blog-post for another day.
More importantly, all of a sudden, all the arguments in defense of raw milk that I have trotted out obediently, on behalf of Neal’s Yard Dairy to officials and other quality assurance managers, clarified in a moment of epiphany.
If you don’t want to bottle milk and offer your customers long shelf life of what is naturally a short shelf life product, pasteurisation is irrelevant.
It doesn’t make sense for a cheesemaker to seek out milk with low counts or be encouraged to use pasteurised milk which it is perceived as being safer than raw milk.
If the raw milk has a healthy population of natural bacteria, it’s the safer choice.
A lot of what a laboratory scientist may consider to be a risk and that has been worked into HACCP and the food safety risk analysis we all do, suddenly seems misdirected and possibly dangerously so.  Pasteurising doesn’t sterilise.  It doesn’t make milk a perfectly clean slate, there are still some organisms in there or organisms can get in there even if you think you’re doing everything at the very pinnacle of hygiene.  With low counts, it’s key to your cheese quality that your starters work quickly.  If they are slow to start, then they can be out competed and potentially they don’t get the upper hand.  This doesn’t neccessarily happen – you might be lucky but it’s like driving without your seat belt.  It isn’t a given that you’ll have an accident but if you do the consequences are worse.
I now see very clearly, that it’s not a question of absolutes and black and whites.  It’s not that low counts of staph aureus or enteros are automatically good.  It really depends on what they compete against.  Cheesemaking is a question of managing populations and communities of organisms.  It’s so much more complicated, nuanced and subtle than low counts good, high counts bad and it’s not a question of limits and levels but of balances.  This is as necessary for food safety every bit as much as to get the recipe to work.
I thought I understood this when I worked at Neal’s Yard Dairy.  I was only half way there.
Now, I really get it.