Which Glass Holds More?

weirdly accurate.

    Which Glass Holds More?

    What is the glass volume illusion test?

    This glass volume illusion test lets you find out whether the shape of a glass changes which drink you think is bigger. You pick the glass you think holds more, then both are poured into identical cylinders and the water decides, five times over. Researchers have asked the same question with real glasses: in a 2005 BMJ study, Wansink and van Ittersum had 198 students judge and pour drinks in differently shaped glasses, and 86 Philadelphia bartenders pour them.

    The test runs as five quick rounds. Each round shows two drinks in differently shaped 3D glasses (three in the last round); you pick the one you think holds more, then both are poured into identical graduated cylinders and the water decides. The rounds run from easy to hard: the biggest drink’s lead shrinks from about 27% to about 11%. Every volume is computed from the exact shape drawn on screen, not typed in beside it. At the end you get your score out of five, how many times a glass’s shape fooled you and which way, and how much drink your picks left on the table.

    It is not an eye test and it gives no percentile: no study has published how individual people score on a pick-the-bigger-glass task like this one, so the result sets your picks next to what researchers found with real glasses instead of ranking you. Five rounds is a small sample, and one slip moves your score a lot. The bias is not a law of nature either: a 2014 Dutch study found that the kind of pour people were asked for changed the outcome, and the result sheet explains how.

    What the test measures

    One thing: whether the shape of a glass changes which drink you think is bigger.

    The task

    • Five rounds of drinks in differently shaped glasses: tall and slender, short and wide, cone-shaped cocktail glasses
    • Pick the glass you think holds the most, then watch both poured into identical graduated cylinders
    • Easy to hard: the biggest drink leads by about 27% in round one and about 11% in round five, which adds a third glass
    • It tests the elongation bias that Raghubir and Krishna described in 1999; the result sheet explains what it is once you finish

    Your result

    • Your score out of five, and how many times a glass’s shape fooled you, and which way
    • How much drink your picks left on the table, in millilitres
    • Reference points from the research: 198 students and 86 Philadelphia bartenders in the 2005 BMJ study, and a 2014 Dutch study of pouring
    • No percentile, because no published study gives individual scores on this task
    • A share card and a challenge link; the link carries only which rounds you got right, in the part after the #

    Before you start

    • About a minute: five picks, one per round
    • Works on a phone or a computer; sound is optional
    • Free, with no account; your browser needs 3D (WebGL) switched on

    Glass shape and volume FAQ

    Is this test free? Do I need to sign up to see my result?

    Yes, it’s free, and there’s nothing to sign up for: no account, no email, no payment. Your result appears on this page the moment you finish, and you can share it or play again straight away.

    Do tall glasses hold more than short, wide glasses?

    Every round of this test asks a version of that question, so the page lets you pick first and the water answers straight after: both drinks are poured into identical graduated cylinders, with every volume computed from the exact shape drawn on screen. Once you finish, the result sheet explains how height and width each count in a glass’s volume, and what people judged with real glasses.

    What did researchers ask people to do with real glasses?

    In Wansink and van Ittersum’s 2005 BMJ study, 198 students estimated how much two 355 mL glasses held and poured a shot into each, and 86 Philadelphia bartenders, with about six years’ experience on average, poured into tall highballs and short tumblers, some of them told to take their time. Their results appear on the result sheet after your five picks, so they can’t steer them.

    Does the glass illusion always work the same way?

    No, and that is part of what makes it interesting. In Caljouw and van Wijck’s 2014 study with ordinary-sized glasses, the kind of pour people were asked for, a generous glass of lemonade or a shot, changed the outcome. The result sheet shows how, after your five picks, along with what decides where the eye settles on a glass.

    How is the glass test scored, and why is there no percentile?

    Each round scores one point if you pick the glass that really holds the most, computed from the exact shape on screen (1 cm³ = 1 mL). The result also counts how often a taller drink or a wider rim fooled you. There is no percentile because no study has published how individuals spread on a pick-the-bigger-glass task, and the site does not publish its own players’ data.

    How hard does the glass test get?

    The glass test runs from easy to hard. In round one the biggest drink leads by about 27%; by round five it leads by only about 11%, and a third glass joins the line-up. The shapes range from tall and slender to short and wide to cone-shaped cocktail glasses, and every volume is computed from the exact shape drawn on screen (1 cm³ = 1 mL).

    Is the which-glass-holds-more test an eye test?

    No. The glass test shows how everyone’s eyes judge volume, not how well you see. It takes about a minute, works on a phone or a computer, and needs no account. Five rounds is a small sample, so a single wrong pick moves the score a lot.

    Where the numbers come from

    The study figures come from real glasses, estimates and pouring, a different task from picking the bigger glass on a screen, so the result compares directions only and gives no percentile. Other papers from Brian Wansink’s lab were retracted in 2018–19; no retraction notice was found for the 2005 BMJ paper used here. The geometry on the result sheet is textbook maths. The site does not publish its own players’ data.