Experience shows that fluids participating in the nature of water rarefy with an enormous force when they are freezing in intense cold.

Ice indeed floats on water. Ice thus is lighter than water. Therefore, it is rarer than water itself. This is deduced from the principle of Archimedes.

At the Accademia del Cimento we proved by countless experiments that ice occupies more volume than fluid water. They are all recorded in the Book of Experiments from page 127 through page 165 where the process of freezing of fresh water is dealt with, resulting either from the artificial cold of snow or from the cold of natural air. In artificial freezing it is always true that, at the beginning of the immersion of a glass vessel ABD in snow RSTV sprinkled with salt, water at first suddenly jumps about 3° upwards, from point E, i.e. from degree 142 up to F (Table 11.​3, Fig. 10). Here, although the volume of water in the vessel appears to increase and rarefy, nevertheless I noticed and learnt that this results from a contraction of the glass vessel itself. Afterwards, from point F, the volume of water decreases and condenses continuously until it is depressed down to point G at degree 120. Here it seems to remain immobile for a little while. Then the volume of fluid water begins again to increase and rises from the lowest mark G up to point H, i.e. up to degree 130. A little later the water carries out a very impetuous jump up to degree 166, at I. Precisely then the water contained in the vessel AB becomes cloudy and turns into ice during the small and imperceptible time during which the very quick jump of the water occurs. Moreover, while the ice acquires more hardness and some fluid parts near the extremity AC of the neck freeze, a flux of water proceeds upwards from the mark I towards D where water brims over and leaves the vessel. This process (without mentioning countless other experiments) obviously shows that water when freezing rarefies, i.e. it expands and its volume increases. The same is observed in water falling drop by drop, in thermal waters, in wine, in vinegar, in lemon juice and in spirit of vitriol. Only air, spirit of wine, oils and quicksilver escape this common law. Their volume decreases and contracts more and more as a result of a more intense degree of cold. Although oil acquires some consistency and solidity, air, spirit of wine and quicksilver always remain fluid.

The force which dilates water in the process of freezing is huge. It appears from experiments recorded at the Accademia del Cimento that glass vessels closed everywhere break when the enclosed water is freezing. And even copper vessels also closed break even if their walls have the thickness of half the auricular finger. This could not be achieved by the force and energy of a wedge compressed by a heavy weight.

Some of the modern authors attempt at explaining the cause of rarefaction of ice, firstly by the principles of Gassendi who expressly denies that cold is pure deprivation of warmth. But, as in nature there are igneous corpuscles producing heat, there are also some tetrahedronal corpuscles which Gassendi calls frigorific or salnitral. These, when introduced into water, are thought to create the expansion of the volume, its binding and its hardness together with immense cold. These authors think that this is the cause of the rarefaction or expansion acquired by frozen water.

The rarefaction and increase in volume of frozen water do not occur by addition and interposition of corpuscles creating cold.

This, it seems to me, can be refuted in two ways. Firstly, the salts added to water should increase its weight and gravity somewhat. But this is against experience. A cup of water of 1 lb for example, weighed on a very precise balance, does not acquire a new weight after being frozen. Add to this that a considerable mass of salt is required to freeze the same mass of water so much expanded because the salt would have to be dispersed through all the particles of water so as to create a universal union and condensation. Since salts by their nature are heavier than water, the weight should increase considerably in frozen water. Claiming that these particles of salt are volatile is not good enough. Indeed, observations carried out at the Accademia del Cimento show that a volatile salt is not different in substance, consistency and shape from a fixed salt of the same kind.

Moreover, if water changed into ice increases in volume because of some frigorific or saline bodies are introduced into and mixed with its own substance, of course all fluid bodies ought to increase in volume and expand like frozen water under the action of the same degree of cold. By hypothesis bodies are made cold only because they are filled and impregnated by these frigorific corpuscles or salts. But this is not true. Indeed, air, spirit of wine, oil and quicksilver even exposed to a North wind do not increase in volume. They even condense and diminish. If they are exposed to a more intense degree of cold than that which is sufficient to freeze water, they maintain their fluidity and decrease more and more in volume, i.e. they occupy less and less space. Consequently, rarefaction and augmentation of volume of ice do not result from sprinkling and mixing of frigorific corpuscles and salts. But this phenomenon results from a by far different cause.

This Gassendi seems to have announced when he says that many particles of air are mixed in ice because we see that freezing of water begins at the top, in contact with air, and from there propagates downwards towards the bottom. Ice thus floats on water because it is impregnated by air, he says. He likely was convinced that the volume of frozen water can be increased by this air. This is conjectured from his own words when he says: Since it is true that hot water when cooling freezes more quickly and more strongly than cold water, do we think that there is another cause than, the laxity of the parts of water being made greater, air is introduced more easily and binds more strongly the particles of water with which it is mixed?

From these words of Gassendi it comes out that in the process of freezing water is inflated and rarefied by air coming from outside.

The volume of water is increased neither by new air arriving from outside in the process of freezing nor by the straightening and tension of the eels of water.

If this happened, when entry of air into water is prevented absolutely, the volume of water should not rarefy and expand when freezing. When a leaden or golden vessel full of water and closed tightly is exposed to very cold air or immersed in snow mixed with salt, the entry of air into water would be absolutely impeded and prevented by the consistency of the metal. Therefore, in the process of freezing water should not rarefy nor expand in volume. But this is against experience. Actually, the leaden or golden phial, thanks to its softness, yields to the expansion of the ice inside and swells, making a sphere of a longer diameter. Moreover, a little before freezing, bubbles rise from the depths of the water in the vessel. They do not appear to sink into the water of the vessel from the outer air. Consequently, frozen water cannot be rarefied and inflated by new air penetrating its substance.

Nor does it rarefy as a result of the straightening of the small eels composing water, as Descartes thinks. This opinion indeed appears to be unlikely, firstly because of the absurdity of the concept. Water indeed cannot be made of these small eels, as we suggested above. Moreover, countless bubbles appear anew which were inconspicuous previously. Therefore, it must be admitted that water is not inflated and rarefied by the straightening and tension of small eels but by air or by another concomitant cause. For the solution of this problem, it must be noticed that countless particles of air are always found mixed with and disseminated in fluid water. This occurs either because the water bordering air, in its varied agitation, intercepts and retains in itself some particles of air, or because countless particles of air are displaced from the subjacent earth, exuding through its pores together with igneous exhalations, and are introduced into the mass of water. If they constitute a great volume, they aggregate and form bubbles which are carried continuously from the bottom of the water up to its surface. This is observed at random at the sea-shore at the places where the ground is muddy, especially in the summer when the sea is quiet. Then many bubbles are seen rising from the bottom at different places. But, whatever the cause of this addition, it is most obvious that a considerable abundance of particles of air is found in water itself even if they are not conspicuous. This can be confirmed with the most beautiful instrument of Torricelli in which vacuum is produced by way of water. Indeed, while water descends to the usual depression of about 17 cubits, we see such an abundance of bubbles getting out of the water that they form the same ebullition as that which is usually provoked in the water by the heat of a fire. This results from the fact that the very small particles of air are no longer compressed by the immense weight of air of the region, as they were previously, but only by the small gravity of the incumbent water. This is suggested by the observation that the deeper bubbles which were almost inconspicuous because of their smallness expand and inflate the more the closer they arrive to the top. They constitute bigger and bigger bubbles some of which are as big as nuts in so far as the freed elastic force of air can expand and dilate these bubbles. Thus it is certain that countless imperceptible particles of air are contained in water, sometimes more, sometimes less abundant. It is no wonder that light air can be retained in a heavier fluid since it is not new that different corpuscles, very heavy as well as very light ones, are retained in water and remain quiescent as a result of the minuteness of their volume, as was suggested above.