Cocktail Science: Does Crushed Ice Dilute More?

posted by Dave Arnold     

A while back we did a post on whether or not the type of ice you use in a shaker affects the dilution and temperature of your finished cocktail. Turns out it doesn’t (most of the time). For the nitty-gritty see here and here. Most people say, “Yeah, I know about your results — but we all know that crappy ice dilutes drinks more. If you shake a drink with crushed ice it will get watery.”
This is both true and false. Here are my thoughts:

  1. Bar ice is at 0 degrees C (just believe it).  Therefore,
  2. All chilling is through melting (read our two previous posts, links above, if you need proof).
  3. Fact 2 is the single most important thing to remember about shaking. Because all chilling happens through melting, for a given amount of chilling there will be a given amount of melting. There is a one-to-one relationship. As long as the surface area of the ice is large enough –and the agitation of the shaking is great enough (which it usually is)– the size or type of ice shouldn’t matter much. BUT….

  4. The smaller the individual pieces of ice are, the greater their surface area per gram. 
  5. Because the ice is at 0 degrees C it is wet – water is on the surface.
  6. The greater the surface area, the more water on the surface of the ice. So….
  7. Shaking with small ice makes the drink watery right away.  The ice brings its own water.  After the initial dilution, small ice and big ice will behave identically (with respect to dilution and temperature).

About two months ago, Don Lee from Momofuku Ssam bar and John Deragon, formerly of PDT, came in to test this hypothesis. Unfortunately, I lost all the photos (don’t ask).  About a month ago we re-ran the experiment and got the same results. Not a hundred percent scientific but here it is:  

Hypothesis:  The extra dilution bartenders experience when using crushed ice comes from water adhering to the surface of the ice.

Test: Dry off some crushed ice and whole ice, shake them both with booze and measure the alcohol content.   

Balancing liquor to the tenth gram.

We measured equal amounts of 901 tequila… Justin Timberlake’s tequila.  The tequila that is bringing sexy back (can we get more free crap yet).   

Crushed ice.

We put ice in a bag and crushed the heck out of it with a rolling-pin.    

Spinning ice in a centrifuge.

We put the crushed ice and and some whole ice in separate bags. We cut holes in the bags for drainage and put them in a centrifuge over drainage racks.  We lightly and quickly spun the bags to throw off excess water.  The centrifuge is one hell of a salad spinner.     

With great celerity we measured equal weights of crushed and whole ice and shook them with the tequila at the same time for the identical number of shakes.  We strained them at the same time through a chinois, to remove the effect of ice crystals.  

Shaking side by side.


  Both drinks had the same volume and both registered 21.5% ABV.   

Standard Ice on left, crushed ice on right. Same liquid volume.

Then we did the same experiment with crushed ice and whole ice that hadn’t been spun in the centrifuge.  The whole ice came in at 21% ABV, slightly more diluted than the spin-dried whole ice.  The crushed ice came in at 20% ABV, a lot more diluted than the spin-dried crushed ice.   


Some Calculations: Let’s assume that all ice comes in perfect cubes.  A single cube 5x5x5 centimeters has a surface area of  150 square centimeters and weighs 115 grams (ice has a density of 0.92g/cc).  It would take 125 ice cubes measuring 1x1x1 centimeters to have the same weight.  Those ice cubes would have a surface area of 750 square centimeters. It would take 1000 ice cubes half a centimeter on a side to have the same weight and they would have a surface area of 1500 square centimeters.  The surface area adds up pretty quickly –so does the entrained water.

19 thoughts on “Cocktail Science: Does Crushed Ice Dilute More?

    1. I don’t have them. They weren’t measured absolutely, they were measured against each-other. The liquor pours were roughly 75 mls, but they were within a 10th gram of each-other (we used a trip balance). The ice weights were roughly 225 grams but also within a tenth of each other.

      1. What’s interesting is that the increase in dilution isn’t nearly as large as you would expect it to be given the magnitude of the increase in surface area.

        If you started out with 75 ml of 40% booze, you had (more or less) 30 ml ethanol and 45 ml water.

        The spun-dry crushed ice came out at 21.5% alcohol, which means 30 ml of ethanol and 109.5 ml of water (64.5 ml added water from melting)

        The untreated crushed ice came out at 20% alcohol, which means 30 ml of ethanol and 120 ml of water (75 ml of water added from melting).

        The untreated whole ice came out to 21% alcohol, which means 30 ml of ethanol and 112.9 ml of water (67.9 ml of water added from melting).

        That means that the untreated crushed ice had 16.3% more water contributed from the ice (9.5% more water total) compared to the spun-dry crushed ice.

        The difference between the water added by untreated crushed ice and untreated whole ice was only 7.1 ml (around 10% more from the untreated crushed ice). This difference seems small given the huge differences in surface area. I’d be interested to see the difference between spun-dry normal ice and spun-dry crushed ice.

        1. Howdy Slkinsey,
          Yeah, it is weird. I’d like to figure out a better way to do it because the centrifuge is a pain use this way and I always feel I am introducing some innaccuracies. It might be better to crush some ice, leave some ice whole, and actually make them dry by freezing them to identical temperatures below 0. They would both require the same number of joules to heat to 0 C. What do you think?

          1. Yes, that makes the most sense to me. It’s hard to keep crushed ice from clumping up when you (re-) freeze it, but I wouldn’t think that would make much difference for your purposes. As you say, you would be starting from the same baseline, so the results should be exensible to a lower starting temperature condition.

            I can believe that surface water is definitely an important variable (probably the most important one. But crushed ice has a surface area to volume ratio that is orders of magnitude greater than it is for cube ice, and this not only has the ability to contribute more surface water but also offers a much larger surface for thermal transfer and melting. I have to believe that this makes some difference, but it would be interesting to see how much difference it makes. The way to do that would be to eliminate surface water and look only at the difference in dilution that is dependent upon melting.

            Looking at both of them, then, you would have the ability to do some modeling as to how much of the observed differences in dilution is contributed by surface water and how much by increased melting rates. You could look at both of these using a variety of sizes of ice (e.g., Kold-Draft versus Scotsman pellet ice versus fine crushed ice).

            I also think that your dilution (and thus ABV) results would be more accurate by simply measuring the final product in a highly accurate way. A highly accurate mass scale would be good, but a very narrow column (burette) would be just as good and a whole lot less expensive. Going over to this methodology rather than using a refractometer (which are notoriously cranky with respect to calibration, and highly accurate ones are very expensive) would allow you to do these experiments starting with lower ABV samples (e.g., with fruit juices).

  1. How do you all measure the ABV% in the kitchen, and bar? Hydrometer, refractometer etc. etc., Any other books or papers you recommend for further reading?

    1. Howdy Moody,
      We use a refractometer that is calibrated to measure percent alcohol. They are pretty cheap. It only works if there is no sugar in the mix –straight booze. We have a hydrometer as well. Measuring ABV in a sugared drink would be far more difficult. The only cheap way would be to measure the density of the drink, then boil off 2/3 of it, refill to same volume with water and measure density again. That’s a pain. What kind of books are you looking for?

  2. While all of this is very instructive, I find the drying and identical temperature aspect of the experiment a bit strange in that “spun” ice is not likely to be used in a practical setting anytime soon.

    Additionally, while the wet ice used in your experiment may be more prevalent than others various establishments I’ve had the pleasure of working in keep the shaking, rocks and hi-ball ice in a forced air freezer well below zero Fahrenheit while the crushed ice sits in an insulated drained bin.

    Though I’ve no experimental data to support my suspicion, I’d be willing to bet the internal temperature of single massive chunks of ice kept in a freezer, even when “wet” from sitting out is well below the surface temperature and the freezing point of water.

    Any thoughts of conducting this experiment in situ? (We’d happily host) After all, we don’t shake in a vacuum.


    Joseph Schwartz
    Little Branch

    1. Howdy Senor Schwartz,

      You are 100% correct that the spinning isn’t a practical suggestion. Its just meant to show where the actual dilution is happening. It isn’t the smallness of the ice that is killing you, its the large surface area and entrained water. The more important information to be gleaned is that medium sized cubes (crap ice) doesn’t have that much more entrained water than larger cubes, like Kold Draft. It also might be useful to give your ice a shake or two before you use it if you are unfortunate enough to use shell ice that stores water like a cup.

      Keeping ice in the freezer, as you know, keeps surface water down as long as you don’t have problems with cracking (I know you guys are experts on ice tempering).

      Ice is a good conductor of heat. It is 4 times better than water. Additionally, ice has a lower heat capacity than water, so it doesn’t provide that much cooling before it starts melting (of course if you use a lot of ice that is a lot colder than freezing you could accomplish a lot of chilling prior to melting). I think the center of the ice will come up to freezing relatively quickly.

      I’d love to do the experiments on with you. We’ll just freeze some cubes with a probe in them, pull them out and see how long they take to go to 0 (32). We’ll do a couple of sizes.

  3. On a separate note, have any other effects of varying ice been noted? For example, aeration, formation of a fine crystalline froth, other textural, visual and presentation aspects that make up some the enjoyable qualities of a well shaken drink?

    Thanks for your efforts at uncovering some aspects of the mechanics of what goes on in the shaker.


  4. I don’t think enough has been made of the amount of ice used and it’s relation to the overall temperature of the vessel i.e. a shaker i/2 full of ice will create more dilution than a shaker full of ice because t it stays colder – despite the fact there is more surface area in a full shaker of ice.

    1. Howdy Jason,
      All of our tests so far show that too little ice results in underchilled watery drinks. I think it is because they aren’t efficient enough to chill rapidly. I’m not sure yet. Our tests also show that after a certain point, adding more ice doesn’t help.

    1. Hi Erik,
      I haven’t done the work, but my feeling is that stirring is just a less effective form of shaking (from a heat transfer standpoint). The basic premise of all the cocktail posts is that there is no chilling other than through the melting of ice.

      1. I suppose that is true.

        Still kind of curious, as there is a significant difference between the contact of the ice/drink with a shaken tin and that it experiences when stirred in a heavy tempered glass vessel.

        For example, many New York bars chill the glasses the use for stirred drinks in freezers.

        I can imagine that the heavy chilled glass might have a more significant chilling effect than the ice. Heck, the ice might even warm the drink.

        Also curious about the data when stirring with cracked vs. cubed ice.

        1. Hi Erik,
          I’m sure Eben Klemm will have us doing the data for stiring soon. I think we are doing it at this year’s tales of the cocktail. Chilling glasses will certainly help.

Comments are closed.