ApiaryActiveLive
Try: pause · settings · learn · wipe
← Community / Reading Room
TZ
craft · 24 min read

Two Zeros on One Shoreline: How to Read a Tide Gauge Beside a Land Elevation Map

A waterfront exhibit wants to show a tide level next to a map of the surrounding town. The chart says 1.34 meters. The land tile says 1.21 meters. The draft…

A waterfront exhibit wants to show a tide level next to a map of the surrounding town. The chart says 1.34 meters. The land tile says 1.21 meters. The draft caption reads, “The water stood thirteen centimeters above the path.” The numbers are real to two decimal places; the subtraction might still be meaningless.

The difficulty is not arithmetic. Each provider chose a reference surface for its vertical numbers. Water and land systems routinely use different surfaces. Two instruments can report in meters, correctly, while measuring from different zeros. A responsible comparison begins by naming those zeros and ends by saying what the comparison can and cannot establish.

This is a guide to evaluating a public map and improving its documentation. It is not a method for deciding if a route is safe in current conditions, determining a legal flood designation, setting a building elevation or making an insurance decision. Those uses require the appropriate authorities, current warnings and qualified professional judgment. An educational demonstration can explain why a color changes on a map without pretending that the color authorizes anything on the ground.

Ask what each number measures

A water level can be an observation, a prediction, a scenario, or a datum value in a table. A prediction describes an expected tide under its modeling assumptions, not a measurement already made. A real-time reading belongs to an instrument with a location, time and datum. A mapped scenario may show what a chosen increase relative to a reference surface could mean under specific assumptions. A datum value can describe a local statistical reference, not the level at the moment of a photograph.

The NOAA Coastal Inundation Dashboard describes the use of local mean higher high water as a common reference for interpreting inundation. NOAA's tidal-datum material and technical manual explain, among other examples, that mean higher high water averages the higher of the daily high waters over the applicable datum period, while mean lower low water uses the lower of the daily low waters. These are local statistical surfaces, not universal elevations with the same difference along every shore.

A terrain value has its own type. Is it a point extracted from a bare-earth digital elevation model? Is it a contour at a published interval? Is it the finished top of a boardwalk, a value labeled by a surveyed benchmark, or a surface model that might include structures? The USGS overview of DEM reference systems gives typical choices but notes exceptions. Check the actual product rather than copying the agency's usual datum into a blank field.

Record quantity type before the number. “Observed water height relative to station MLLW” and “bare-earth terrain cell relative to NAVD88” are data capable of being reconciled if further conditions are met. “Water: 1.34” and “park: 1.21” are not. The underlying datasets may be excellent; the loose note is what prevents someone from checking your conclusion.

Separate the vertical frame from the horizontal one

Map software makes showing two layers together feel like the work is done. A web map can automatically display datasets in the same horizontal map view while doing nothing about vertical datums. The marker sits exactly beside the color tile. It does not follow that the two vertical values are comparable, or even that the land pixel sits where you think it does when consulted at its native scale.

Identify both horizontal and vertical reference information. Keep location, horizontal reference where relevant, vertical datum, unit, resolution, source and date as separate fields. At a sharp waterfront edge, a slightly different location can sample water, fill, the top of retaining structures or adjacent soil. The sensitivity is not the same as a vertical reference error. Changing one cannot fix the other.

Locality matters vertically as well. The difference between two tidal surfaces can vary with the local water body and with location. A conversion for a harbor station cannot simply be copied into a town across a bridge. Nor does a transformation magically transfer observed conditions to the path, even if you can express both measurements on the same datum. If the gauge is nearby rather than co-located, instrument location, water connectivity, tide timing and local conditions still influence what you can conclude.

Think ahead to a future editor. They should be able to see why a vertical transformation was appropriate at the tested coordinate, whether that coordinate represented the gauge or the terrain location, and why any observed reading was reasonable to show beside the terrain. Two different icons can share a diagram without claiming the same local event.

Make the reference shift explicit

You can often explain the needed operation with a picture of two horizontal lines. Consider a fictional location where the MLLW zero sits 0.30 meters below the NAVD88 zero. A water level said to be 1.34 meters above the MLLW zero is then 1.04 meters above the NAVD88 zero in this illustration. The shift in this case is negative because the original zero is lower in the destination frame. In another hypothetical situation the sign could differ. Save both the formal transformation direction and the meaning of the sign, then check the calculation against the stated relationship between zeros.

For real U.S. coastal work, NOAA VDatum provides documented transformation tools and uncertainty information. Verify actual location coverage, acceptable coordinate and datum definitions, and the reported or documented limitations for that model region and transformation path. Save the model or tool version when available and the day you retrieved the result. If the location is outside a documented domain, mark the conversion unsupported; do not extrapolate an offset from a convenient station simply to make the comparison possible.

Use the input and target datum names exactly. Saying “from sea level” obscures whether a source means local mean sea level, a chart datum, a broad terrain reporting reference, or something informal. The source metadata matters even if two offsets happen to be close in one location. Record what you did rather than naming the assumption after the result you wanted.

A helpful quality check is to transform a demonstration value of zero in the source frame wherever the documented tool and workflow support the exact same transformation. The outcome gives a cross-check for the sign and comparison in the specific setup, not an invitation to generalize the shift across a map. If a colleague switches input/output fields and gets a superficially plausible answer, this simple reference test is easier to diagnose than a complicated elevation claim.

Worked example: exhibit path or map pixel?

Everything about Breakwater Park in this example is invented. The values are chosen to illustrate the arithmetic and the effects of missing metadata, not to describe a real station, map tile or boardwalk.

The design team has a station value of 1.34 m relative to MLLW and a broad land raster value of 1.21 m referenced to NAVD88. It wants to compare both in a single figure. For teaching purposes only, suppose a compatible transformation at the relevant example location puts source zero at −0.30 m in NAVD88, and suppose the illustration treats water height from the nearby station as applicable to the point. Both premises need a real verification record in practice.

OperationDemonstration resultInterpretation
Raw numbers: 1.34 − 1.21+0.13 mNot a valid height comparison; different reference surfaces
Convert water height to terrain frame: 1.34 + (−0.30)1.04 m NAVD88Correct under this fictional transformation and units
Display common-frame comparison: 1.04 − 1.21−0.17 mSurface is modeled below the selected terrain cell in the illustration
State what was measuredSelected terrain cell; not path top or a certified site heightDo not say the path was dry

The change in sign catches the editor's attention. But a common reference is not the same as a valid claim. Imagine that the selected terrain cell is on the downslope side of an elevated path, or that a parent display rounded a contour into a spot number. A terrain product intended to model bare ground may not represent the deck. The right replacement caption could be “Comparing two vertical frames changes the relationship between the example water surface and the selected terrain cell; the figure does not measure an actual path.” That's still an interesting result because it shows how a common mistake can reverse a visual story.

Make the lesson more robust with a second check. Pick two plausible neighboring cells in the synthetic illustration: 1.13 m and 1.42 m. The mapped comparison against water at 1.04 m is now −0.09 m and −0.38 m respectively. Nothing changed about the water datum. Sampling changed the land value. If changing a pixel is enough to change the practical interpretation, enlarging the display or printing more decimal places is going to look impressive without adding evidence.

Now replace the observation with a scenario from a viewer that shows water as an increment over local MHHW. Use that scenario's definition rather than applying a station MLLW offset from the previous exercise. The map viewer is not showing the same quantity as the momentary station observation. Before even considering conversion, the new comparison requires a new caption explaining the scenario and its assumptions. An exhibit can include both if it keeps the labels and the question each answers distinct.

Read accuracy, resolution and freshness as separate signs

A one-meter terrain cell is a sample spacing or product grid characteristic, not a claim of one-meter vertical error. Nor does a height returned with many decimal places imply the underlying measurement is that precise. The USGS 3DEP vertical-accuracy FAQ reports accuracy estimates for specific evaluated products against checkpoints; the 3DEP spatial metadata guide points to work-unit boundaries and, where available, project-level materials. Read through to the applicable local metadata before assigning an error figure. Don't borrow a national aggregate as a guaranteed local bound.

Vertical conversion can add uncertainty, and the measured input and output are not always independent. Identify the relevant components and their provenance. Is the terrain number a product-level error estimate? Is the water value a checked observation, a preliminary reading or a scenario? Does the model report uncertainty for the exact conversion indicated? Don't add numbers with unlike meanings or call a generic sensitivity example a statistical interval. A small gap between the converted water and mapped ground may not be resolved by those inputs, even if a plotting library assigns different colors with no hesitation.

Age affects the question in a different way. A terrain tile acquired before a major ring of landscaping may be perfectly valid for that earlier surface while omitting a later slope or path. Record collection year separately from the portal's refresh or access time. A historic station series can belong to a documented reference period different from the one in a new product. Version the comparison rather than assuming both state the coast as it is now. In an educational plot show the intended period, and if timing doesn't line up, say so.

Don't turn a threshold into a complete inundation model

Reconciliation answers one question: can two elevations be expressed in a comparable reference frame under the stated assumptions? It doesn't explain how water reaches a place, where barriers redirect it, whether drains affect low terrain, how waves act, or what an official warning means. A low cell surrounded by higher ground presents a different mapping problem from a low cell connected to open water. An embankment narrower than a grid cell might be missing or generalized.

The NOAA Sea Level Rise Viewer describes its role as a preliminary visualization and documents limitations, including aspects of future terrain change, levees and elevation data. If you use a view from it in a figure, identify the viewer, variable and chosen scenario. Keep the original map credits and caveats attached. A screenshot colored by an educational condition should not be relabeled as a route recommendation, a current mapping alert or a substitute for official determinations.

Use an editorial uncertainty category when the audience needs a simple description. For example, “adequately bounded for this illustration,” “difference not well resolved,” and “datum unverified” are outcomes about the evidence. They are not classes of real-world safety. Put the explanatory text near the visual rather than in a remote appendix most viewers will never see.

A quick editorial exercise: find the missing field

Show a second writer the two original values and the proposed headline, but no captions. Ask them to name the first piece of metadata they would request. If they ask only for extra precision, show a pair of diagrams in which each scale measures from a different line. If they ask whether the data came from the same place, use that as an opening to separate spatial transfer from reference conversion. A strong exhibit gives readers enough information to notice both problems, not just enough to memorize three initials.

For a reproducibility check, imagine the exhibit has to be rebuilt without scripts next year. Write a short alt-text equivalent: name the two zeros, explain that one line is lower than the other in the fictional model, and report the direction of the corrected comparison without indicating actual coastal depth. This is a useful discipline for the chart maker. If the reasoning is impossible to describe without detailed mouseover behavior, the chart probably still has an unlabeled assumption.

Record which date belongs to which observation

A water level can have an observation timestamp, a prediction can have a valid time, a tidal statistic can describe a reference period, and a terrain layer can have an acquisition interval. A data service's download time belongs to a fifth category. Don't put all five into a spreadsheet column called date. A new download can retrieve an old photograph or an earlier numerical forecast, and having two files acquired on the same afternoon doesn't make their represented observations simultaneous.

For a display comparing land height with a historical water event, capture the time basis as locally published and the offset or named convention where available. If a station output says UTC, retain it as UTC rather than making it match the local photograph by eye. A photo caption might record the local time as remembered rather than as encoded by a camera. Where the source does not supply enough information to align the two, do not invent an interval. A timestamp is important only to the claim you are making; a generic sketch of the relation among vertical zeros doesn't need an event time, while a claim that a gauge and a photo show the same event does.

The NOAA bench mark information guide for observing systems describes networks of physical marks used to reference and recover local measurement heights. Here the useful distinction is between a station's working reference, the statistical datum and a reference to a mark on land. A physical mark can play a role in tying these frames together; it is not sufficient to see a number stamped on metal and enter it into a conversion as an elevation. Find the station or mark identifiers and current relevant information for whatever limited educational use the program has in mind. This article does not teach how to occupy control marks or perform a survey.

Check numbers in actual station information rather than applying an assumed version of a datum across epochs. NOAA's datum update overview describes the recomputation work associated with changing the National Tidal Datum Epoch. The useful workflow is to log the epoch labeled on the input and the datum in the output, and keep the referenced documentation. It would be a mistake to relabel all historical inputs with the latest name just because screen headers changed. If labels seem identical but epoch or transformation version differs, show why they are comparable or leave a break in the graphic. No calculation can restore an unrecorded version number.

Separate three attractive overlays that answer different questions

A public poster can show a water-level timeline, a terrain profile, and a map of a modeled scenario. These are three distinct products. The timeline shows values at one station over time under some reference. The profile places land or surface values along a route under its own frame and sampling. The modeled map depicts spatial results under assumptions. Which one belongs on the poster depends on the sentence the image has to support.

If the intended sentence is “the gauge was higher than its long-term high-water reference at this time,” the station series and datum metadata are the main evidence. Do not choose a bright top-down screenshot merely because it looks like the shore. If the sentence is “this set of mapped ground cells has less height than the chosen reference after alignment,” an actual profile or labeled samples will do more honest work. If the sentence is “a model shows a connected pathway by which water can reach this area under these specific assumptions,” preserve the model documentation and scenario settings. Don't say this last thing on the basis of the second alone.

Try a one-sentence legend for each selected overlay. For a station timeline, state the product type, station location and vertical basis. An observed level, astronomical prediction or statistical summary should have a visibly different label. For terrain, state surface type where known, place, reference and date. For a scenario map, identify assumptions about the indicated water level and modeled connectivity rather than borrowing actual occurrence language. If the input documentation cannot supply any of these distinctions, do not let an elaborate color scale suggest precision the authors of the graphic don't have.

The distinction matters especially when the area of interest is behind an obstruction or connected by a channel not represented in the chosen terrain grid. A threshold scheme can mark cells below a height under a uniform plane; it does not explain why water gets to them. Modeling connectivity can require information not represented in a station value, such as operating controls or effects of channels. A viewer built for communication may intentionally simplify these topics. State the simplification. Do not invent local engineering details to fill that void, and don't use a public exhibit to make access, construction or emergency decisions.

A second worked example: two dates at the Bath Lane stair

Bath Lane, its station names, example data and outputs below are fictional. An exhibition about the waterfront shows a photo captioned “Steps wet an hour before predicted high tide.” Beside it the designer wants two values: 1.42 meters above a local tidal zero from a station plot and 1.55 meters on a public terrain product at a location near the stair. The planned callout says “Water was 13 centimeters below the stair.” None of the authoring team has yet checked that either number establishes the thing its label says.

Start with what kind of value each is. The station plot is a prediction rather than a contemporaneous measurement. Its timestamp corresponds to the prediction interval and a location on water some distance from the stair. The photograph is of a wet-looking lower tread and was given a local time by a donor; it cannot explain what wetted the surface. The terrain product is described as ground elevation. It does not claim to represent the built tread. It may be useful for geographic context, but a data cell near a structure is not a physical measurement of that structure. The author already has three independent reasons not to publish the callout even before any arithmetic.

In a fictional file supplied for education, both numeric values can be compared over the area of the schematic after an approved illustrative conversion. The documented demonstration says to add 0.18 meter to the prediction to express it in the terrain frame. The resulting nominal height is 1.60 meters. Compared to the 1.55-meter terrain cell, the numbers are now arranged in the other order. This reversal illustrates what a different origin can do to a difference. It is not a result about any real street, station or stair. State the transform's direction explicitly as “target-frame height at this sample equals source-frame height plus 0.18” before drawing an arrow.

The nominal five-centimeter difference in the demonstration is less than plausible uncertainty in several input components stated in the classroom worksheet. The station prediction is not an observed event, and no validated transfer of water level to the stair exists. No amount of rounding can make that small nominal difference answer the actual question. The scientifically interesting lesson in the exhibit may be the change in the arithmetic under a common reference, not the illusion of reconstructing a photograph from two near-looking values. Show a middle caption: “With the example data aligned, the numbers are close; the inputs don't tell us where the water reached the stair.” That outcome lets visitors learn about coordinate systems without relying on an invented waterline.

What would strengthen a narrower claim in a real research setting? A well-dated photograph locating a visible mark, a contemporaneous reading at an appropriate station in a verified common frame, and an independently documented surface of interest might improve context. A validated local inundation assessment, not this exercise, would be needed for consequential conclusions. Finding one better input can change one question without changing the others. Replacing the prediction with a measured station series doesn't convert a ground sample into a tread height. The record should make these independent failure modes visible.

Build a range, not a decimal race

When the inputs can be compared for a limited demonstration, keep a small uncertainty inventory. Terrain accuracy, unresolved surface type, transform uncertainty and methods of estimating a water value are different contributors. Some are numerical uncertainties reported in metadata. Others are categorical limitations, like not knowing if a raster represents a tread at all. A numerical range cannot compensate for a wrong physical object. Start by excluding mismatched objects, then decide what remaining uncertainty is meaningful to report.

In a fictional range-only exercise, a converted representative water value is between 1.48 and 1.62 meters under stated assumptions, while an appropriate terrain profile estimates a particular surface between 1.52 and 1.65 meters under another set of stated assumptions. The intervals overlap from 1.52 to 1.62 meters. Neither number is always greater across the ranges. This does not mean the two observations are equivalent or that all conditions in the overlap were equally likely. It means that a strict ordering is not supported by those illustrative bounds alone. The minimal useful classroom display shows two differently labeled bands, not a pair of crisp lines separated by one centimeter.

Don't add errors until they look like a rigorous estimate. Limits with different statistical meanings and correlated inputs aren't a simple uniform population. NOAA's VDatum uncertainty documentation explains uncertainty drivers and separately discusses model components and source information. It is sensible to preserve units, method, confidence language where supplied, and location-specific caveats without pretending those are uniform. If combining statistical quantities is outside the project's expertise, report them separately and say what conclusion could not be determined. A small group teaching datums is not obliged to publish a numerical propagation it cannot justify.

Use a context checkpoint when converting throughout a map. Ask whether the point lies within documented coverage, whether the tested transformation is permitted for those frames and that place, and whether unusual conditions or documented limitations apply. A value returned by tooling, if available, is not the same as verification that those inputs properly describe the scenario. Conversely a gap in the model doesn't authorize interpolating across a headland from the last plausible values. Mark an unsupported portion of a demonstration as untransformed and give the reader a plain explanation. If a graphic is mainly to show the meaning of two zeros, one committed, clearly fictional example can teach more than an almost comprehensive map with hidden extrapolation.

Review the numbers from outside the GIS

A handoff test can be done without loading the whole mapping project. Write the statement the public graphic makes, the station product and its original reference and time basis, the terrain source and surface, the operation that connects the frames where supported, and one limiting fact. Include original numbers before rounding and label which scale the rounded display uses. Ask another reader to reproduce the direction of the conversion and say what changes if the sign is wrong. Then ask whether the values allow the claim. These are distinct tests; a reader might calculate a correct subtraction while still comparing an inappropriate object.

Give the visualization a legend that is accurate when detached from the article. A screenshot might travel without accompanying prose. Words such as “instructional scenario, not an observation” can be essential in the graphic itself. A carefully selected map frame with one station dot and one site label might tempt viewers to interpret the profile as a proven link between them. Draw connections only to explain what was actually transferred or compared. A visible annotation describing the station as a regional reference but not as a site sensor is better than relying on the reader to notice a footnote.

For a public map describing only shoreline context, retain the actual project date on the legend. Mapping products and station metadata may be revised. If a source update changes any input, keep an old calculation note and add a new comparison rather than overwriting the original as if never shown. The new version should separate new surface observations from changes to the origin or method. Readers should be able to tell whether a differently colored bank is a newly mapped landform or simply a choice of zero. Never use an apparently higher-resolution screenshot to imply a substance change you haven't traced to a revised dataset.

One tempting editorial exercise is to present a table with units replaced by blanks and ask participants which missing field they would request first. Repeating the exercise at higher difficulty, use two labeled numbers in meters but omit the named station on one row and the type of surface on the other. Readers who immediately ask for a datum have half of the method; a second question about the actual object being measured makes it stronger. A correction record that catches mislabeled objects before a map is drawn saves far more time than updating a glossy display after people start pointing at it.

Consider making the uncertainty itself part of the learning objective. A map of observation points and a short scenario diagram can show where evidence stops, such as where an estimate is valid only at a named station rather than across a bay. A visitor can then distinguish an open inquiry from an untrustworthy chart. Publishing a gap with an explanation is neither a failure of mathematics nor a temporary embarrassment. It preserves the data's meaning, which matters more than finishing a smooth line in a poster.

Separate counts of events from counts of samples

An educational chart sometimes shifts from heights to the number of water-level events. Even after the vertical reference is aligned, a count depends on what an event means. In an hourly series a single pulse may produce five samples above a threshold. In a daily summary the same pulse may show up as one day above a threshold. Neither is the same thing as five independent events. Decide whether the display shows samples, dates, or episodes and say how runs are grouped. If the chart repurposes an agency event count, preserve the agency's actual definition instead of giving the count a familiar label that means something else.

Data gaps can interact with grouping. In an invented three-hour span around a peak, imagine a value above the operating threshold of the exercise, then a missing sample, then another value above. Did the episode continue or were there two separate exceedances? The samples don't establish either answer without a convention and enough context. A classroom rule could leave the episode continuity unresolved and report two positive samples. That reports exactly what was observed in the invented sequence without inventing a crossing between them. A chart counting events can annotate this ambiguous interval; do not turn a missing observation into a level of zero to force a complete line.

The same caveat applies to comparing sampling products through time. If the first series records one value per hour and the second uses daily maxima, both can have units of elevation and still make incompatible event views. An aligned datum is a prerequisite for a height comparison, not a cure for mismatched sampling. Write the statistic, interval, station and time convention next to the vertical reference. If a table contains candidate values of different kinds, one short notes column can keep a volunteer from inadvertently combining them because a pivot table accepted the whole column.

Even an apparently robust same-station count cannot certify how frequently a particular stair was wet. It establishes an observation pattern under a narrow definition at the station, missing data included. That distinction is a useful capstone for the poster: measuring a sequence, choosing a zero and deciding what a count represents are three separate choices. A reader should know which choices were actually made before being shown a dramatic rising line.

Give the next editor a minimal build sheet

A useful build sheet for an exhibit summarizes the question in one line. Then it records the following:

  • Water source, station and coordinate; observation, prediction or scenario; associated timestamp, units and reference.
  • Terrain source, product, acquisition date, coordinate, vertical frame, cell size and corresponding accuracy metadata.
  • Transformation input/output, location, supported model coverage and round-trip or zero check as applicable.
  • Numerical calculation and distinction between fictional arithmetic, actual raster evaluations and verified measurements.
  • Local limitations, including whether surface of interest, observation location or connection assumptions remain unresolved.
  • Final caption, date reviewed, and authoritative sources for any decision the educational exhibit cannot support.

Test the handoff by deleting the chart from a temporary copy. Could another editor reproduce the educational comparison from the sheet, and would they know not to infer local safety? Check that both numbers in the title callout name units and reference, or label them in an obvious shared legend. Write whether your final figure used source units or converted them, and keep the original values so someone can detect a future conversion error without starting from scratch.

Sometimes the honest final output is a pair of uncombined public layers and a caption about why a link between them is missing. That is a useful map-reading lesson. Two solid sources that answer different questions don't have to be dragged into a single precise-looking number. In coastal displays, one of the most useful things you can label is exactly where the zeros came from.

AI disclosure: Drafted with AI assistance for Apiary. The Breakwater Park example, dimensions and offsets are synthetic, not station observations or an assessment of a real site. Educational mapping only.

Source notes

Frequently asked
What is Two Zeros on One Shoreline: How to Read a Tide Gauge Beside a Land Elevation Map about?
A waterfront exhibit wants to show a tide level next to a map of the surrounding town. The chart says 1.34 meters. The land tile says 1.21 meters. The draft…
What should you know about ask what each number measures?
A water level can be an observation, a prediction, a scenario, or a datum value in a table. A prediction describes an expected tide under its modeling assumptions, not a measurement already made. A real-time reading belongs to an instrument with a location, time and datum. A mapped scenario may show what a chosen…
What should you know about separate the vertical frame from the horizontal one?
Map software makes showing two layers together feel like the work is done. A web map can automatically display datasets in the same horizontal map view while doing nothing about vertical datums. The marker sits exactly beside the color tile. It does not follow that the two vertical values are comparable, or even that…
What should you know about make the reference shift explicit?
You can often explain the needed operation with a picture of two horizontal lines. Consider a fictional location where the MLLW zero sits 0.30 meters below the NAVD88 zero. A water level said to be 1.34 meters above the MLLW zero is then 1.04 meters above the NAVD88 zero in this illustration. The shift in this case…
Worked example: exhibit path or map pixel?
Everything about Breakwater Park in this example is invented. The values are chosen to illustrate the arithmetic and the effects of missing metadata, not to describe a real station, map tile or boardwalk.
References & sources
  1. Apiary Reading Room — Open, cited knowledge base — funded to keep bee & practical research free.
From the Apiary Reading Room. Opinion & editorial — not financial advice. We don't overclaim.
More from the Reading Room