Answers to the self-checks

We are looking for the name of a passage from one state to another, and for that we use the table of the six changes of state, then the distinction between the two vaporisations. At the start, the water soaks the cloth: it is in the liquid state. At the end, the towel is dry and nobody has wrung it out or heated it: the water left in a form the eye cannot see, so in the gaseous state. The passage from liquid to gaseous is a vaporisation. It is not a boiling for two reasons, and each one is enough on its own: boiling water calls for one hundred degrees under the usual pressure whereas the room is at eighteen degrees, and a boiling takes place throughout the whole body of the liquid, making bubbles, which did not happen. The answer is therefore a vaporisation, and more precisely an evaporation, which takes place at any temperature, slowly, at the free surface of the liquid.

We are looking to settle a claim, and for that we use the property established in the chapter: water vapour is perfectly transparent. A camera records only what interacts with light; a transparent gas cannot therefore appear white in an image. What the photograph shows is something else: water vapour did indeed come out of the spout, but it occupies the perfectly transparent zone just above it, and it became visible only after meeting the cooler air of the room and condensing back into tiny liquid droplets. The answer to give is therefore: that white plume is liquid water, already condensed back, and the real water vapour is in the transparent zone that comes before it, where there is precisely nothing to photograph.

We are looking for observable differences, that is to say ones that can be seen without an instrument, and for that we use the section devoted to the two vaporisations. First difference, the temperature: evaporation takes place at any temperature, boiling at a fixed temperature, one hundred degrees for pure water under the usual pressure at sea level. Second difference, the place: evaporation only takes place at the free surface of the liquid, boiling throughout its whole body. Third difference, the bubbles: boiling produces them, evaporation produces none. Two other differences are acceptable if they replace one of the previous ones: the speed, slow against fast, and the noise, silent against noisy.

We are looking for the name of a passage, and for that we use the table of the six changes of state together with the fact that air always holds invisible water vapour. The layer keeps its shape and scratches under the fingernail: it is in the solid state. The question rules out rain and states that the windscreen was dry, so no liquid water can have settled and then frozen; the only source of water available is the vapour in the air. The surface having cooled below zero degrees during the clear night, that vapour settles directly in the solid state without passing through the liquid state. The answer is therefore a deposition, that is to say a direct passage from the gaseous state to the solid state, and it is that direct passage which gives hoar frost its needle-like look instead of a smooth sheet of ice.

We are looking both for a prediction and for the limit of what an experiment allows one to conclude, and for that we use the factors that speed up evaporation together with the rule of experiment 1: only one thing must change from one run to the other. The plate in the sun will empty first, because the radiation warms it and because warmer water, with warmer air above it, evaporates faster. But this experiment does not vary a single factor: in the sun the water is warmer, the air above it is too, and the draught may differ. The only honest conclusion is therefore that the whole set of conditions in the sun speeds up evaporation compared with the shade. One cannot say which of the factors is responsible, and above all one cannot conclude that sunshine is needed for evaporation to take place, since the plate in the shade empties as well.

We are looking to compare two quantities of water from two percentages, and for that we use the definition of relative humidity as the ratio between the quantity of vapour contained and the maximum quantity possible at the temperature of the moment. The trap consists in comparing 90 and 55 directly. Now these two percentages do not relate to the same maximum: maximum capacity rises with temperature, so the air at thirty degrees can hold far more vapour than the air at three degrees, and 55 % of a large maximum is well above 90 % of a small one. The answer is therefore that the summer air at thirty degrees carries the more water, and that the percentage never compares two bodies of air with each other: it compares each body of air with its own limit of the moment. To compare two bodies of air, it is the dew point that must be looked at.

We are looking to state a condition and then apply it, and for that we use the section devoted to the single condition of condensation. The condition is stated thus: there is condensation when the air is cooled to its dew point temperature, that is to say to saturation. Application to the mist of a breath: air breathed out leaves the lungs warm and practically saturated with water vapour; on contact with the cold outside air, it is abruptly cooled below its dew point, and the surplus it can no longer keep in the gaseous state condenses into tiny droplets which become visible. So one never sees the breath itself, which is a transparent gas: one sees the water it was carrying and which has just become liquid again. That also explains why the phenomenon disappears as soon as it turns mild, the outside air no longer being cold enough to bring the breathed-out air to its dew point.

We are looking to explain a distribution, and for that we use the single condition of condensation, looking for what cools and not for what wets. The quantity of water vapour available is the same everywhere in the garden: it is therefore not that which makes the difference. What differs is the cooling. During a clear night, the grass in the open loses its heat towards the sky and falls well below the temperature of the surrounding air, down to its dew point: water settles on it. The grass under the table is sheltered: the table screens it, it does not lose its heat in the same way, it stays warmer and does not reach the dew point. The answer is therefore that the difference comes from the cooling and not from the water available, and it is exactly the same reasoning as for the mist on the kitchen window, the part of the cold surface being played here by the exposed grass.

We are looking at what a control run is for, and for that we use the rule that an experiment must isolate the ingredient responsible. In the full run, two things are brought together: air loaded with vapour, and a cooling from above. Without the control run, one would only show that bringing the two together makes a white veil appear, which is true but does not say which of the two is at work: somebody could maintain that it is the hot water alone that makes the fog. The control run removes one single ingredient, the cold, and leaves everything else identical; since nothing then appears, it establishes that it is indeed the cooling that sets the phenomenon off. The answer is therefore: without a control run one would show that a result exists, but not what causes it, and an experiment that does not separate its ingredients allows no conclusion about a mechanism.

We are looking to compare two bodies of air on their readiness to give fog, and for that we use the definition of the dew point as the temperature at which a body of air reaches saturation. Both bodies of air are at fifteen degrees, so their starting temperature does not separate them. The first has a dew point of fourteen degrees: it need only lose one degree to become saturated. The second has a dew point of two degrees: it must lose thirteen. A night does cool the air, but it cools it by a certain amount only, which depends on the season and on the cloud cover. The answer is therefore the first body of air, the one whose dew point is fourteen degrees, because even a modest night-time cooling is enough to bring it to saturation, whereas the second would need a far greater cooling to give fog.

We are looking for the explanation of an apparent contradiction, and for that we use the section on the size of the droplets and the movements of the air. The droplets of a cloud are so small that the air slows them down as much as they weigh: they do come down, but so slowly that the least upward movement is enough to hold them up or to carry them back up. Now the air of a cloud is precisely air on its way up, since it is that rise which cooled it and which brought the cloud into being. The answer therefore holds in two sentences: it is not that the droplets do not fall, it is that they do not fall fast enough to leave the cloud; and the upward movement that created the cloud is also the one that keeps it in the air.

We are looking to name a mechanism, to justify a difference between two surfaces, then to cite the other three, and for that we use the list of the four ways of cooling air. Cauliflower-shaped clouds are cumulus, and they are born of convection: the air present near a warm ground is warmed on contact with it, expands, becomes lighter, rises and cools at height until it condenses. The difference between the land and the lake comes from the surface: on a summer afternoon, land warms up quickly and strongly under the Sun, whereas a body of water warms up much more slowly and stays cooler; above the lake there is therefore no warm ground to start convection off, and no column of air rises. The three other ways of cooling air are orographic lifting, when a body of air pushed by the wind is forced to rise along a relief; frontal lifting, when warmer air slides up over colder air at the boundary between two bodies of air; and cooling from below, when mild air moves over a colder surface. Remember that the first three make the air RISE and that the fourth does not: it is the ground or the sea that takes its heat away.

We are looking to read a name instead of reciting it, and for that we use the Latin roots given in the chapter. Altostratus breaks down into alto, which marks the middle level, and stratus, which means a spread-out layer: it is therefore a layer of the middle level, and one can deduce from that a greyish sheet covering the sky at an intermediate height. Cumulonimbus breaks down into cumulus, which means a heap, and nimbus, which means rain: it is therefore a heaped cloud that gives rain, and one can deduce from that a rounded mass able to produce precipitation. The expected answer is that breakdown, plus the remark that makes it useful: these roots combine freely, which makes it possible to understand a name never met before without having learnt it.

We are looking for the rule of classification, and for that we use the definition of the level given in the chapter. A cloud is sorted by the height of its base, and never by that of its top. Now the base of a cumulonimbus is low, often less than a kilometre above the ground, which places it in the low level, defined as running from the surface of the globe to two kilometres. That its top rises to fifteen kilometres and so crosses all three levels changes nothing about the classification. The answer is therefore: because the classification is made on the base and not on the top, and that convention exists precisely so that an observer on the ground, who sees the base and not the top, can classify a cloud without an instrument.

We are looking to finish a piece of reasoning begun in the worked example, and for that we use orographic lifting together with the link between temperature and maximum capacity for vapour. On the side where the wind pushes the air against the slope, the air is forced to rise, it cools, it reaches its dew point and it condenses: that is the wet slope, the one that receives the clouds and the precipitation. Past the crest, the air comes back down the other slope, it warms, its maximum capacity to hold vapour rises again, and the droplets evaporate once more: that is the dry slope. The complete answer is therefore: the wet slope is the one facing the sea wind, the dry slope is the opposite one, and the cloud breaks up past the crest because the descent warms the air instead of cooling it.

We are looking for what separates a cloud that keeps its water from a cloud that lets it fall, and for that we use the section on the two roads to growing. What has changed is neither the nature of the cloud nor the quantity of water it holds: it is the size of its droplets. They have clumped together, through successive encounters or by passing through the solid state at height, and Météo-France puts it thus: when they become too heavy, they fall as rain. That is enough because a larger droplet comes down faster: beyond a certain point, its speed of fall exceeds that of the upward current which was carrying it, and it leaves the cloud from below. The answer is therefore: only the size has changed, and it is enough because it is what decides whether the fall wins over the rise. As for the forms that what falls can take, there are five. Drizzle, liquid, made of droplets whose diameter does not exceed 0.5 mm. Rain, liquid, whose drops exceed 0.5 mm: that single threshold is what separates them. Snow, solid, which forms when the temperature of the air is below or close to zero degrees. Ice pellets, solid, made of ice grains finer and softer than hailstones, and which no diameter defines at the source. Hail, solid, made of ice particles of diameter greater than 5 mm. The squally shower, for its part, is not one more form but a mixture: a brief shower mixing rain, hail, ice pellets, snow and sleet.

We are looking for a volume of water, and for that we use a single relation, that of Météo-France: one millimetre of rain is equivalent to one litre of water per square metre. It fits exactly, because it links a depth in millimetres to a volume per unit of surface, and because the question gives a depth and a surface. The surface of the roof is 9 times 6, that is 54 square metres. A depth of 8 mm is equivalent to 8 litres per square metre. The total volume is therefore 8 times 54, that is 432 litres. Check of the order of magnitude: 432 litres is a little more than two bathtubs, which is plausible for moderate rain on a roof of that size. The answer is 432 litres, that is about 0.43 cubic metres.

We are looking for an argument, and for that we use the comparison between two containers of different sizes. If a volume were measured, the result would depend on the container: under the same shower, a large bucket collects far more than a glass, and the two readings would not be comparable. A depth, for its part, does not depend on the size of the container: under the same rain, a bucket and a straight-sided glass fill to exactly the same depth, because the surface that collects and the surface that receives grow together. The answer is therefore: a depth is measured because it is the only quantity that does not depend on the instrument used, which makes two readings comparable wherever they are taken; and it is from that independence that the equivalence between a millimetre and a litre per square metre follows.

We are looking for measurement errors and their direction, and for that we use the rules for siting a rain gauge. First error, shelter: the overhang of the roof intercepts part of the rain before it reaches the container, and the reading is then lower than reality. Second error, runoff: water collected by the roof can drip along the overhang or the wall and fall into the container, and the reading is then far higher than reality, since it counts water that fell on a surface far larger than the opening of the rain gauge. The answer is therefore those two errors, one by default and the other by excess, with the remark that matters: they can occur together without cancelling out, which makes the reading not only false but impossible to correct afterwards. As for the weather radar, it collects nothing: it sends out a wave and measures what the drops send back to it, over a range of about 100 km for measurement, with an image every 5 minutes. It does not do away with the need for a rain gauge for two reasons: it sees rain UP IN THE AIR, sometimes before it touches the ground and sometimes when it will evaporate before arriving, and it measures indirectly where the rain gauge collects for real. The two instruments answer two different questions, one about what is falling on a large scale, the other about what actually arrived at one point.

We are looking for the reason behind a measuring practice, and for that we use the trap of the chapter about millimetres of rain and centimetres of snow. A depth of snow cover does not measure a quantity of water: it measures the way the flakes have piled up. In hard frost, the same quantity of water gives a thick, light cover; at a temperature close to zero, it gives a thin, heavy one. Two snow readings in centimetres therefore cannot be compared with each other, and nor can they be added to rainfall readings. By melting the snow it collects, the service brings everything back to the same quantity, the depth of liquid water. The answer is therefore: so that its readings are comparable with each other and with those of rain, which is the very condition for a measurement to be a piece of data.

We are looking for the property that makes a material permeable, and for that we use the section devoted to permeability. The answer lies neither in hardness nor in appearance, since both materials are loose. What counts are the voids the material contains, and above all the fact that those voids connect with one another. Sand is made of grains that leave spaces between them connected to one another: the water goes down from one to the next, so it is permeable. Clay is made of tiny flat particles that stick to one another: its voids exist but they are too fine and too poorly connected to let water through at any useful rate, so it is impermeable. The answer is therefore: it is the size of the voids and their connection that decide, and not the hardness of the material.

We are looking to describe a journey using three imposed words, and for that we use the description of infiltration given by Eaufrance. First sentence: after the rain, the drop enters the pores and the cracks of the soil and it is drawn downwards by gravity. Second sentence: it thus passes through the subsoil until it meets an impermeable bedrock, which interrupts its descent. Third sentence: it then builds up in the empty spaces of the porous or cracked rock lying above that bedrock, that is to say in the aquifer, and the water so accumulated makes up the groundwater. The answer is that journey in three stages, and the point not to be missed is that the aquifer means the rock while the groundwater means the water it holds.

We are looking to separate what a question allows one to conclude from what it does not, and for that we use the list of runoff factors together with the notion of saturation. What the question allows one to settle: in BOTH cases, heavy runoff is possible, and the intuition that would point straight to the thunderstorm is too hasty. What it does not allow one to settle: which of the two runs off the more, since the exact intensity of the storm and the degree of saturation of the ground are missing. The factors at work in the first case: the rain is fine, so not intense, which works FOR infiltration, but the ground is already sodden, so saturated, its voids are full and the water arriving has nowhere to go, exactly as if the ground were impermeable. The factors in the second case: the ground is dry, so its voids are available, which works FOR infiltration, but the storm is brief and intense, so it brings water faster than the ground takes it in. What must be remembered is that the intensity of the rain is not the only factor, and that the prior state of the ground weighs just as much.

We are looking for two consequences, one at the surface and the other underground, and for that we use the list of runoff factors together with the role of infiltration. At the surface: tarmac is impermeable, so almost all the rain runs off instead of soaking in, it arrives faster and in greater quantity in the networks and the watercourses, and the risk of overflow during an intense shower rises. Underground: the water that no longer soaks in no longer joins the groundwater, so the underground reserve is recharged less, which will be paid for later, in a dry spell, by a smaller discharge to the watercourses. The answer is that pair, and the remark that links them: the quantity of rain received has not changed, only its sharing between surface and depth has, and that is what makes both the floods and the low flows worse.

We are looking for a conclusion and then a calculation, and for that we use the runoff factors and then the relation between millimetres and litres per square metre. Conclusion: the clay hillside runs off the more, since its three characteristics all point the same way, impermeable ground, steep slope and bare soil, whereas the flat, grassy sandy plateau favours infiltration on all three of its characteristics; the intensity of the rain nevertheless causes some runoff on both. Calculation: 25 mm is equivalent to 25 litres per square metre, and one hectare is 10,000 square metres, so the storm laid down 25 times 10,000, that is 250,000 litres, in other words 250 cubic metres. The answer is therefore: the clay hillside runs off the more, and the storm brought 250,000 litres per hectare.

We are looking for a definition and for the argument that makes it work, and for that we use the definitions given by Eaufrance. A catchment is the territory corresponding to the whole of the surface receiving the waters that flow naturally towards one and the same watercourse or towards one and the same body of underground water. Its boundary is the watershed, a natural boundary drawn by the relief and corresponding to the ridge lines: the drops falling on one side or the other of that line feed two neighbouring catchments. It does not follow administrative borders because it was not decided: it is traced by the relief, which takes no account of human divisions. The answer must contain the consequence: one district can belong to two catchments, one catchment can cover several countries, and that is why water is managed by catchment.

We are looking for explanations consistent with an apparently contradictory fact, and for that we use the table of the three inputs of a watercourse. First explanation: the input from tributaries. A shower that fell upstream, sometimes several tens of kilometres away, takes several hours or several days to come down, and it makes the river rise when not a drop has fallen locally. Second explanation, and it is of another kind: direct runoff on a sealed-over sector of the catchment. A shower that fell on a built-up area, a car park or a bare clay hillside, even a nearby one, let almost nothing soak in; all the water left along the surface, very fast, and it arrived within a few hours although no drop fell on the point of observation. The answer is that pair, and the point of method it illustrates: what arrives in a river does not depend on what falls where we are looking at it, but on what falls over the whole of its catchment and on the way each sector of that catchment deals with what it receives.

We are looking for the origin of the water that flows outside spells of rain, and for that we use the section on the three inputs. Runoff has long been exhausted, since it acts within a few hours. The tributaries can bring nothing more, since they are in the same situation. That leaves the third input: the discharge from groundwater. Water that soaked in weeks or months earlier comes back out slowly into the bed of the watercourse and makes up what is called the baseflow. The answer is therefore: it is the discharge from groundwater that makes the river flow in August, and that water comes from the rain of previous seasons, the rain that soaked in instead of running off. That is the exact sense of the formula in the chapter: what the ground took in during the winter is what will flow during the summer. As for recognising it in the field, that is done without any instrument and on three signs. In low flow: the water is CLEAR, because it comes mainly from groundwater and not from the washing of soils; the level is low and uncovers gravel banks; the bed shrinks to a narrow channel. In flood: the water is BROWN, because it is carrying particles torn from the soil; it is high and fast; and it sweeps branches along. Colour is the surest of the three signs, because it says where the water comes from.

We are looking to explain an apparent paradox, and for that we use the relation between infiltration and baseflow. Building over replaces permeable surfaces with impermeable ones: tarmac, concrete, roofs. The rain falling on them no longer soaks in, it runs off. Immediate consequence: it arrives faster and in greater quantity in the watercourse, so the floods are stronger and more abrupt. Delayed consequence: that same water has not recharged the groundwater, so the underground discharge will be smaller in the following months, so the low flow will be more severe. The answer is therefore that the quantity of water received has not changed, but that its spread over time has got worse: everything arrives quickly and nothing is put into reserve, which produces both drawbacks instead of one.

We are looking to confirm a surprising claim, and for that we use the definition of the watershed. Yes, they can. The top of a ridge is a watershed, that is to say a natural boundary between two neighbouring catchments. A drop that fell on one side runs off towards the outlet of its catchment, the other towards the outlet of the neighbouring catchment, and nothing will make them meet since the two catchments overlap nowhere. If the two catchments empty into different seas, the two drops end up in two different seas. The answer must add the qualification made in the chapter: part of each drop soaks in rather than running off, but that does not change the arrival, because groundwater too flows towards the outlet of its own catchment.

We are looking for an ordered list together with the changes of state, and for that we use the section that runs through the loop. Evaporation of the oceans, the rivers and the lakes under the effect of the heat of the Sun: liquid to gaseous. Evapotranspiration, which adds to that vapour the transpiration of plants: liquid to gaseous as well. Formation of clouds, when the vapour rises and turns into tiny droplets or ice crystals: gaseous to liquid, or gaseous to solid. Precipitation, when the droplets become too heavy and fall as rain, snow or hail: no change of state for rain. Sharing at ground level, finally, between runoff towards the rivers and the large rivers, build-up in glaciers and infiltration towards groundwater: freezing has already taken place at height for snow and hail, melting comes in when they melt at ground level or when a glacier retreats, and evaporation takes over for the share that goes straight back into the air. The loop closes when the large rivers rejoin the ocean. As for the great reservoirs, there are six: the oceans and the seas, the atmosphere, the glaciers and ice sheets, groundwater, watercourses and lakes, and finally living things and soils. By far the most is held by the ocean, since water covers nearly three quarters of the surface of the globe and that is where the bulk of the reserve lies.

We are looking to answer a question and to show that it is badly put, and for that we use the property that defines a cycle. Answer to give: the water cycle has no beginning, because each of its stages is at once the consequence of the previous one and the cause of the next, endlessly; asking where it begins makes no more sense than asking where a circle begins. Why the question is badly put: it imports the structure of a chain, which has a first stage and a last, into an object that is a cycle and has neither. One may add that the order in which a book sets the cycle out is a convenience of narration, and that beginning with a cloud, with a body of groundwater or with a puddle would give exactly the same sequence, shifted by one step.

We are looking for an explanation that can be deduced, and for that we use the fact established in chapter 1 that a change of state concerns only the water itself. When sea water evaporates, what passes into the gaseous state is the water; the salt, for its part, does not evaporate and stays in the ocean. The vapour that rises is therefore water without salt, the clouds it forms are not salty, and the rain that falls from them is no more so. The answer is therefore: the water cycle works like an immense separation kept going by the Sun, which draws fresh water out of a salty reserve, and it is that mechanism which makes life possible on the continents. It can be checked at home: after a plate of tap water has evaporated completely, a faint deposit is left at the bottom, that is to say what could not leave into the air.

We are looking to compare two apparently identical acts, and for that we use the trap of the chapter about nested loops of unequal lengths. The cycle is not a single loop: it is an endless set of loops of very different lengths. Water drawn from a river belongs to a short loop, fed by runoff and by discharge from groundwater, and it is replaced within a few days to a few weeks. Water drawn from deep groundwater belongs to a long loop, whose renewal time can be counted in centuries. The answer is therefore: the two abstractions take the same volume but not the same resource, one drawing on a flow that rebuilds itself quickly and the other on a stock that does not rebuild itself on the scale of a human life.

We are looking to finish a journey begun in the worked example, and for that we use the three inputs of a watercourse seen in chapter 6. Mechanism of departure from the groundwater: discharge, that is to say the very slow flow of the groundwater following the slope of its bedrock, until it comes back out either at a spring or straight into the bed of a watercourse. Reservoirs crossed next: the watercourse, then possibly several successive watercourses as far as the outlet of the catchment, and finally the ocean. Changes of state: none over that whole stretch, the water remaining liquid from the groundwater to the ocean; one may note that part of it evaporates on the way from the surface of the watercourse, which sends it back into the atmosphere by a shorter loop. The loop is then closed, since the ocean is the reservoir of departure.

We are looking for an ordered list, and for that we use the table of the small cycle. One, abstraction: raw water is taken from groundwater, a lake or a river. Two, making it drinkable: it is filtered to remove the particles in suspension, then disinfected to destroy the micro-organisms. Three, storage: the drinking water is held in reserve up high, most often in a water tower. Four, distribution: the water is carried under pressure through a network of mains to houses, farms and industrial sites. Five, collection: after use, the waste water is gathered by the sewers or directed to an individual system. Six, treatment: it is cleaned in a wastewater treatment plant or in an on-site sanitation system, then given back to the natural environment.

We are looking to correct a confusion and to describe two installations, and for that we use the sentence from Eaufrance according to which drinking water never comes from wastewater treatment plants. Correction to be made: these are two separate chains, with two different works, two different networks and two directions, and they meet nowhere. The drinking-water treatment plant stands at the beginning of the small cycle: it receives raw water taken from the natural environment and it releases drinking water into the distribution network. The wastewater treatment plant stands at the end: it receives the waste water collected and it releases cleaned water into a watercourse or into the sea, together with sludge. The answer is that pair, and the formula that anchors it: one takes from the environment to give to the tap, the other takes from the tap to give back to the environment.

We are looking to settle a question of drinkability from an appearance, and for that we use the sentence from the Artois-Picardie water agency according to which water that has been pumped up is clean but it is not drinkable. No, it cannot be drunk. Clean and drinkable are not synonyms: clearness tells us only about the absence of particles in suspension, whereas water can be perfectly clear and hold micro-organisms or dissolved substances that no eye can detect. That is precisely why making water drinkable involves two operations and not one: filtration, which deals with what is in suspension, and disinfection, which deals with what filtration cannot reach. The answer adds what this volume repeats every time: drinkability is a matter of regulated checks made in a laboratory, and water is drunk because it is supplied as drinking water, never because it looks clean.

We are looking to avoid a hasty conclusion drawn from two percentages, and for that we use the distinction between abstracted and consumed. What has to be added is that abstracting is not consuming. Water abstracted to cool a power station is largely given back to the environment shortly afterwards, a little warmer; it therefore remains available downstream. The report states that only between 12 % and 15 % of the fresh water abstracted in France is actually consumed, that is to say not given back to the environment. The answer is therefore: the 45 % for cooling measure a volume that passes through, and not a volume that disappears, so that comparing that figure with the 19 % for drinking water without this clarification amounts to adding two different quantities together. Account must also be taken of the place and the moment of abstraction, since pressure on water is always local and seasonal.

We are looking to explain a choice of siting, and for that we use the sentence in the chapter according to which the height of a water tower puts the network under pressure. The height serves to distribute by gravity alone: the water comes down from the reservoir to the houses without a pump needing to run continuously, and the network stays under pressure even if the pumping fails, for as long as there is water left in the reservoir. If it were built in the lower part of the village, the difference in height with the houses further up would be nil or negative: the water would not rise by itself, and a pump running continuously would be needed to supply them, with supply stopping at once at the slightest breakdown. The answer is therefore: the height is what manufactures the pressure, and it is also what manufactures the independence of the network.

We are looking for conditions and, for each one, the error it repairs, and for that we use the sentence from Météo-France about measuring temperature. 1.5 metres above the ground: this repairs the error due to the ground, which warms and cools faster than the air and which would make any measurement at ground level systematically more extreme. In a screen: this repairs the error due to radiation, from the Sun as well as from what the ground sends back, and it also protects the sensor from precipitation. A white screen: this repairs the error due to absorption, a dark surface warming under light whereas a white surface sends it back. A louvred screen: this repairs the error due to confinement, a closed screen ending up measuring the air of its own box instead of the surrounding air. The answer must contain the general principle: a condition of measurement is never a formality, each one repairs a precise way of going wrong. As for the other quantities a station records, they go in pairs with their instrument: the depth of precipitation, by rain gauge, in millimetres; the wind direction, by wind vane, named by where the wind comes from; the wind speed, by anemometer, always averaged over a stated length of time; the relative humidity, by hygrometer, as a percentage, and in the same screen as the thermometer since it depends on the temperature; the atmospheric pressure, by barometer, in hectopascals, reduced to sea level so as to stay comparable; and the cloud cover, which has no instrument at all: it is estimated by eye, in eighths of sky covered.

We are looking to apply a convention, and for that we use the rule that a wind is named by the direction it comes from. An east wind comes from the east and heads west. The face the wind strikes is therefore the EAST face, the one that faces it; the sheltered face is the opposite one, that is to say the WEST face. The point of method to remember is that a mistake about the convention reverses the answer exactly, which makes it one of the few errors that common sense cannot catch: the two answers are symmetrical and seem equally plausible. That is why the convention is learnt instead of guessed.

We are looking to make two different measurements consistent, and for that we use the fact that a wind measurement is always an average over a length of time. Météo-France distinguishes the instantaneous wind, measured over three seconds, from the mean wind, calculated over ten minutes. The two stations are therefore not measuring the same thing: the one announcing 90 km/h is very probably giving a gust, that is to say an instantaneous value, and the one announcing 45 km/h is giving a mean wind. Both can be accurate at the same time, since a gust is by definition a brief maximum within a flow whose average is lower. The answer is therefore: neither is wrong, they bear on different lengths of time, and comparing an instantaneous value with an average makes no sense if the length of time is not stated.

We are looking to separate two neighbouring notions, and for that we use the distinction between the precision of the instrument and the accuracy of the measurement. A very precise thermometer placed in full sun measures its own temperature, which depends on the radiation it receives: it therefore gives a value several degrees away from the temperature of the air, with a fine apparent precision. A rough thermometer correctly screened does measure the temperature of the air, to within a degree. The answer is that precision describes the fineness of reading of the instrument whereas accuracy describes the distance from the true value, and that it is accuracy that counts. The practical consequence is the one given in the chapter: for a record kept at home, look after the location and the regularity first, and only then the instrument.

We are looking to mark out what a series allows, and for that we use the condition of regularity of a record. What you can still conclude: each value stays true at the instant it was taken, and you can still say what the lowest and the highest temperature you met were. What you can no longer conclude: anything comparing days with one another, neither a trend nor an interpretable average, since a difference between two days may come from the weather or simply from the time of the reading, temperature varying a great deal over a day. The answer is therefore: the series becomes a collection of isolated observations instead of a comparable series, and what has to be done is to fix a time and start the series again, never to correct it after the event.

We are looking to describe a complete chain, and for that we use the two previous chapters. The thermometer placed in a white, louvred screen 1.5 metres above the ground produces an accurate measurement, therefore one comparable with that of every other station in the world, since the conditions are standardised by the World Meteorological Organization. That measurement joins the other observations coming from automatic stations, from radars and from dozens of satellites, and out of the whole set the calculation draws a picture of the atmosphere at the instant it begins: that is the initial state, an approximate reconstruction and never reality itself. That initial state is given to a numerical forecast model, ARPEGE or AROME, which works out the likely development of the atmosphere, and the supercomputers Belenos and Taranis run that calculation. Human expertise finally analyses those complex results and turns them into concrete information, that is to say into lines of forecast.

We are looking for a mechanical explanation, and for that we use the fact that the initial state is an approximate reconstruction and not reality. The model starts from a picture of the atmosphere carrying very small differences, which cannot be avoided: between two stations an estimate has to be made, over the oceans measurements are rarer, and some quantities are measured less well than others. Those small differences do not stay small: at each step of the calculation, they spread and grow. After a few hours they are still negligible, which explains why a forecast at twenty-four hours is excellent; after several days they have grown to the point of changing the result. The answer is therefore: the forecast does not deteriorate because the atmosphere would be more complicated in seven days' time, but because the difference at the start has had seven days in which to grow.

We are looking to interpret a change of forecast, and for that we use the fact that each calculation starts again from more recent observations. What has happened is simple: a new calculation was run from a more recent, and therefore better, initial state, and it produced a different result. No, one should not conclude from it that the service is wrong: a forecast that corrects itself is not a forecast that had been wrong, it is a forecast that has received better data. The answer must add the point of reading that follows from it: the closer the range comes, the more reliable the forecast, and that is why the right practice is to look again as the date approaches rather than sticking to the first announcement one heard.

We are looking to read two colours and to say where to find what should be done, and for that we use the table of the four colours together with the safeguard stated in the chapter. Yellow signals: be attentive, that is to say that an event is possible and that the department is named together with the event concerned. Green signals: no particular vigilance, that is to say that the department is not concerned by any event signalled for the period covered. Where to find what should be done: with the map itself, from the service that publishes it, at the moment the danger exists. The answer must contain the reason: what should be done depends on the event, the place, the hour and each person's situation, and no book written months earlier can replace that. It is the general rule of this volume: a safety instruction is read where it is published and kept up to date.

We are looking to word a complete answer and to justify a moment, and for that we use the way reliability falls with range. Answer in two sentences: today, one can say that a regime milder and wetter than the normal is expected for that period, which makes unsettled weather more likely than dry and cold weather; on the other hand, knowing whether it will rain at midday that day goes beyond what a forecast at nine days can give. Moment at which to look again: THE DAY BEFORE, because the chapter establishes that a forecast at twenty-four hours is excellent, the differences in the initial state not having had time to grow. The justification is the one given in the chapter: the reliability of a forecast depends first on the distance between the moment of the calculation and the moment forecast, and that is why one must look again as the date approaches rather than sticking to the first announcement.

We are looking for a single criterion, and for that we use the definition of the two words. The criterion is the scale of time the statement bears on, and it alone: neither the object nor the method separates weather from climate, since both use the same quantities and the same instruments. A statement bearing on an hour, a day or a few days is weather. A statement bearing on tens of years, or on an average calculated over a long period, is climate. Possible examples, among many others: it will rain tomorrow afternoon over the northern half of the country is a statement about the weather, since it bears on one day; July is on average the warmest month of the year in this region is a statement about the climate, since the words on average point to a statistic over many years.

We are looking for a definition, a length of time and a frequency, and for that we use the section on normals. A climatological normal is a statistical reference representing the climate of a given period, against which measured values are compared in order to say whether a month or a year has been warm, cold, dry or wet. It is calculated over a period of thirty years. It is updated every decade, following the rule of the World Meteorological Organization. The answer may add the reason for choosing thirty years: it is long enough for exceptional years to cancel one another out, and short enough to stay representative of a period.

We are looking for a reason and a consequence, and for that we use the change of reference of 28 June 2022. Reason: Météo-France replaced the period 1981-2010 with the period 1991-2020 because the 1981-2010 normals were no longer representative of the present climate in a context of climate change. Consequence for a departure calculated on the old reference: it is not false, but it can no longer be compared with the departures published today, since it is calculated against an average 0.42 degrees colder; the same month would therefore give a larger departure on the old reference than on the new one. The answer must contain the principle: a normal is not a fixed value of nature, it is an average calculated over a chosen period, and reading a departure requires knowing over which period it was calculated.

We are looking to link two chapters, and for that we use the logical order stated at the end of the volume. Climate is defined by averages, spreads and extremes calculated over tens of years, therefore from long series. Now a series only means something if its values are comparable with one another: measured in the same place, under the same conditions, and in the same way from one end to the other. If the conditions change along the way, a difference observed between the start and the end of the series may come from the climate or simply from the change of conditions, and nothing allows one to settle the matter. The answer is therefore: chapter 9 establishes the standard conditions that make measurements comparable, and without them there would be no usable series, so no normals, so no statement about the climate. We measure under shared conditions, we obtain comparable series, we draw normals from them, and only then can we speak of climate.

We are looking to refute a piece of reasoning, and for that we use the fact that no single value tells us anything about a long average. First sentence: a night of frost bears on a scale of time of one night, whereas a statement about the climate bears on a scale of tens of years, and an observation made at one scale can neither confirm nor contradict a statement made at the other. Second sentence: what tells us about the climate is a long series of comparable measurements from which averages are drawn, and a climate that is warming on average forbids no very cold night, exactly as the mean height of a group forbids nobody in it from being very tall or very short. The answer must stay symmetrical: the same reasoning forbids concluding from a single heatwave day that the climate is warming.