In Part I, we looked at what ice does in a drink. Now comes the fun part: putting a few options side by side. I compared household cubes, a large molded sphere, bagged ice, and whiskey stones to see how quickly they cooled vodka and how much dilution accompanied the cooling.
This is a home experiment, not a controlled laboratory study. I used Tovolo sphere molds with a rigid lower section and a flexible cover because they were easy to fill and unmold. The observations below belong to this equipment and procedure. They should help us ask better questions rather than declare one kind of ice the winner in every drink.
The experimental plan
The plan was to make tray cubes and spheres from the same drinking-quality water, freeze them for at least two days, and compare similar masses in matching glasses. I would add vodka, record temperature every minute for ten minutes without stirring, then separate and weigh the remaining ice and liquid.
The original notes proposed a 1.5-US-fluid-ounce pour, about 44 mL. The procedure below records approximately 47 mL, so there is a small discrepancy in the record. I am preserving that difference rather than silently making the experiment appear more consistent than it was.
The original analysis also introduced a “Cool Factor” using percentage temperature reduction on the Celsius scale. That is not a physically sound universal efficiency measure: Celsius has an arbitrary zero, and a percentage changes if the same temperatures are expressed in Fahrenheit. A better comparison reports temperature changes in °C, the full temperature-versus-time curve, and added water at a matched final temperature.
Making the ice
We need drinking-quality water, clean trays and molds, fresh bagged ice, the reusable chillers, matching glasses, a scale, and suitable temperature probes. Ice becomes an ingredient as it melts, so water quality and clean handling matter.
I started by filling the kettle within its safe fill limit. In this trial, I boiled the water before making the homemade ice. This documents what I did; boiling for ten minutes is not a requirement for making ice from safe drinking water and does not guarantee clear cubes.
The kettle went on the stove, with enough space left to avoid boiling over.
I kept the water boiling for roughly ten minutes in the original trial. There is no need to repeat that step merely to get cold ice.
The loss of water through evaporation was easy to see.
I let the water cool before filling the molds. The original notes mention both 65°C and 70°C. For any repeat, use a temperature permitted by the tray or mold manufacturer and avoid handling hot water unnecessarily; room-temperature water is a practical starting point.
Fill each mold only to its indicated line so the water has room to expand as it freezes.
One thing I like about these molds is that they stack, which helps in a crowded freezer.
Then I filled the regular trays. Tilting a connected tray slightly can help water move between compartments, but a stable surface and a steady pour are more important than any trick.
The trays and molds went into the freezer for at least two days.
Two days later
I set out matching glasses, the scale, and the temperature probes. Matching glasses help reduce differences, but they do not eliminate heat exchange with the room or make every part of the liquid the same temperature.
The homemade cubes and spheres came out of the freezer.
I let them sit briefly before unmolding. This tempering step can reduce cracking, but it also changes the starting condition and allows warming or melting. A more rigorous repeat would standardize the tempering time, remove surface water consistently, and record the starting temperature.
The sphere weighed 123 g; the tray cubes weighed 122 g. Close, but not exactly equal—and worth recording honestly.
I added about 47 mL of vodka and positioned the probes near the side of each glass. There was no stirring. I recorded the reading immediately after pouring, then every minute for ten minutes. In a repeat, the probe should remain immersed without touching glass or ice, and the initial drink temperature and elapsed time should be controlled closely.
After ten minutes, I strained the liquid and weighed the liquid and remaining ice separately. Some liquid clings to ice, so draining time and handling introduce uncertainty into the dilution estimate.
The difference between the samples was already visible, but the photographs are only part of the story.
The melted sphere was particularly interesting to look at. Its shape reflects how it sat in the liquid and where heat reached it.
I repeated the comparison with bagged ice and whiskey stones. A reusable chiller is not directly equivalent to ice by mass: its material and heat capacity differ, and it does not absorb heat by melting into the drink.
Reading the results
The first graph preserves the original temperature record. The point labeled −2 minutes represents the vodka before the test, reported as 23°C. The t = 0 point is the reading after the pour, so some cooling had already happened before the timed curve began.
In these recorded runs, bagged ice brought the initial reading to about 6°C, the tray cubes and stones to about 10°C, and the sphere to about 15°C. By ten minutes, the cube and bagged-ice readings were below 0°C, while the sphere was around 6°C. An alcohol-and-water mixture can remain liquid below water’s freezing point, but these single-point readings should not be mistaken for a fully mixed drink’s exact bulk temperature.
The original summary chart is retained as part of the experiment record, including its percentage-temperature panels and “Cool Factor.” Those quantities should not be used as valid measures of cooling efficiency. The dilution notes also need caution: reported factors of 1.62, 1.97, and 1.93 do not all reconcile with the stated 47 mL pour and the listed added-water volumes. Without the underlying measurement sheet, I cannot responsibly recalculate them into a corrected dataset.
The qualitative pattern is still useful: in this setup, the sphere cooled more slowly and diluted less over the ten-minute observation, while smaller ice cooled more quickly and added more water. The stones added no meltwater. A fair efficiency comparison would measure water added to reach the same final drink temperature, with repeat trials and consistent starting conditions.
The original discussion quoted cooling rates around 1.8°C/min for tray cubes and 1.4°C/min for the sphere over selected intervals. Those slopes do not by themselves rank total performance: the samples began the timed period at different temperatures, and the choice of interval changes the comparison. Read the whole curve rather than one number.
The taste test
The sphere sample tasted cool rather than very cold, with some welcome dilution. The cube and bagged-ice samples were colder but more diluted than I wanted for sipping vodka. The stones chilled without adding water, leaving the alcohol’s sharpness more noticeable. These are my tasting impressions, not a universal preference.
What I took back to the bar
Ordinary cubes remain useful, versatile tools. A large sphere is attractive in a short drink and can suit a slower pace. Bagged ice is convenient for entertaining, while reusable chillers let you add any water separately. None of those observations proves that tray cubes are unsuitable for shaking or that a sphere is always the best choice for whiskey.
The useful lesson is to choose ice for the drink and the way you intend to serve it. Cooling and dilution are linked, and an experiment becomes more useful when we are just as interested in its limitations as its result. Next time: more repeat runs, better-controlled probe placement, and a comparison at the same final temperature.