A spirit level, also called a bubble level, is an instrument that checks a surface’s orientation relative to gravity. A small bubble moves inside a curved liquid-filled vial, settling toward the highest part of the chamber. When the instrument’s intended measuring face is horizontal, a correctly aligned vial places the bubble at its reference marks. Other vials can check vertical or specified inclined positions.

Knowing how to use a spirit level means understanding more than where the bubble sits. Level, flat, straight and square describe different geometric properties. The result also depends on the measuring surface, vial sensitivity, instrument condition and whether dirt or a protrusion prevents proper contact. A centred bubble can be a correct reading of an incorrectly positioned tool.

This guide explains spirit-level physics, vial construction, accuracy markings, reversal checks and slope calculations. Worked examples connect millimetres per metre with angles, while practical observations show how to separate a tilted surface from an instrument error. The focus is measurement literacy and ordinary tool use, not structural certification or a universal installation specification for buildings, machinery or drainage.

The 50-second explanation

A spirit level contains a curved chamber filled mostly with liquid and partly with gas. Gravity makes the denser liquid seek lower positions; the less dense bubble occupies the highest available region. Tilting the chamber changes where that highest region lies, so the bubble moves along the vial.

The vial is mounted in a body with a reference measuring face. When the relationship between that face and the vial is correct, the bubble position indicates the body’s inclination. A line vial checks one direction. A circular vial can indicate tilt in two directions together.

The key question is therefore not simply whether the bubble is centred. Ask which measuring face is touching the object, whether the contact is sound, and which direction is being checked. Those conditions determine what the centred bubble actually establishes.

Level, plumb, flat and square are different claims

Level refers to gravity

A locally level direction is perpendicular to the direction of gravity. A horizontal surface is level in every direction across its plane. A long spirit level usually checks inclination along one axis, so a second orientation is needed before treating a tabletop as level in both directions.

Imagine a board balanced along a horizontal line but tilted sideways. The bubble can centre when the tool lies along the line and move when the tool is rotated across the board. Both observations are consistent. The first checked only one direction, not the entire surface.

The reference is local. Ordinary construction measurements cover distances small enough to treat the relevant horizontal plane as flat. Extending that local approximation across enormous distances would introduce Earth geometry and surveying considerations beyond the task of a handheld level.

Plumb refers to vertical

A plumb direction follows gravity rather than standing perpendicular to it. A vertical vial is mounted so its centred condition corresponds to the tool’s intended face standing vertically. A level with several vials therefore offers several specified orientation checks, not several copies of the same horizontal reading.

A post can appear vertical from one side while leaning in another direction. Check appropriately oriented faces or directions. A single vertical reading should not be turned into a claim about every possible lean of the object.

Flatness and squareness require other comparisons

A sloping ramp can be flat without being level. A wavy board can have a level average orientation while remaining uneven. A straightedge helps reveal gaps and local shape, while a spirit vial reports orientation of the tool resting on its supports.

Square means that specified directions or surfaces meet at a right angle. Two parts can be square to one another while the complete assembly is tilted relative to gravity. A square and a spirit level answer different questions, even when both tools are used during the same job.

Naming the property first prevents unnecessary adjustments. Before changing a shelf, decide whether the concern is its tilt, sag, twist or relationship with another part. Each problem calls for a different observation.

Why the bubble moves uphill

The vial is not simply a straight tube with a bubble trapped at a random location. Its internal measuring surface has deliberate curvature. Along that curve, one region becomes highest for each small inclination of the tool. The bubble settles toward that region while liquid occupies the lower space.

Tilting one end upward therefore moves the bubble toward the raised end. The bubble does not act like a heavy bead rolling downhill. Its lower density relative to the liquid is essential to the direction of motion.

The chamber’s curvature determines how much the bubble moves for a small angular change. A gently curved chamber can produce a larger travel for the same small tilt than a more tightly curved one. This is the beginning of vial sensitivity.

Level Developments’ explanation of vial sensitivity relates curvature, bubble movement and angular response. It also distinguishes sensitivity from accuracy. A highly responsive bubble does not automatically mean the vial has been mounted correctly relative to the measuring face.

Surface condition and liquid properties influence how smoothly the bubble settles. A clear, continuous response makes reading easier. Multiple bubbles, leakage or a visibly damaged vial indicate that the tool needs attention rather than a more determined effort to interpret its marks.

The liquid, chamber and markings

The word spirit refers historically to the liquid-filled design, but the exact fluid is a manufacturing choice. Different products use suitable low-freezing liquids and additives. Do not assume every level contains ordinary drinking alcohol, plain water or the same chemical formulation.

The liquid must remain usable over the intended temperature range, provide readable contrast and behave appropriately against the vial surface. Colour helps visibility; it is not the source of the measuring effect. A green bubble vial and a yellow one do not indicate different fundamental kinds of gravity.

Johnson Level’s discussion of vial fluids and pitch markings describes why the chamber, liquid and marks are engineered together. The marks may indicate a central reference or particular slopes, depending on the model. Their meaning should be read rather than guessed.

Two lines around a bubble are not automatically degree divisions. On many ordinary levels, they guide visual centring. Special models may have additional pitch marks with explicit slope values. A bubble touching a line therefore has no universal numerical meaning across all tools.

The chamber itself may be glass or a suitable transparent polymer. Visibility, impact protection and stable positioning matter more to the user than material appearance alone. Do not open a vial or add liquid to repair it; doing so changes the instrument rather than restoring a known reference.

The body is part of the measuring system

A vial reads the orientation of its mounting. The body connects that mounting to the workpiece. If the body bends or the intended measuring face is damaged, the bubble may no longer describe the contact surface accurately.

Common designs use rigid metal profiles or other suitable structures. Their shape balances stiffness, weight and resistance to handling. End caps can absorb some impacts, but they cannot guarantee that an instrument remains accurate after every fall.

Some measuring faces are machined; others use a controlled coated surface. The relevant requirement is a dependable reference face, not a belief that one visible finish automatically proves accuracy. STABILA’s spirit-level construction guide treats vial mounting and measuring surfaces as a coordinated system.

A long tool spans a larger region and can bridge local irregularities. That can be useful for assessing an overall line, but it can also hide a dip beneath the middle. A short tool samples a smaller region. Length changes the question being sampled, not merely the prestige of the instrument.

Do not use the body as a lever or strike it unless the product is specifically designed for that treatment. A tool can still look serviceable while its relationship between base and vial has changed.

Sensitivity is not accuracy

Sensitivity describes how much the indication changes for a given change in tilt. A highly sensitive vial produces noticeable bubble travel for a small angle. That is useful when detecting small inclinations, provided the support and reading method are stable enough.

Accuracy concerns how well the indication corresponds to the intended reference. A sensitive vial mounted at the wrong angle may give a beautifully repeatable incorrect zero. Making the bubble movement larger does not fix that offset.

Readability is another property. A bright vial with clear marks may be easier to centre consistently than a dim one, even when their nominal sensitivities are similar. Viewing angle, lighting and bubble length also affect how confidently the user locates the centre.

Think of a weighing scale with large digits. The large display improves reading, but it does not prove that the scale was zeroed or calibrated. A spirit level has the same separation between sensing response, reference alignment and human interpretation.

For a practical task, choose enough sensitivity to answer the question without making the reading needlessly unstable. A precision machinist’s level and a general household level serve different conditions. Greater sensitivity demands better support and more careful interpretation, not just a more expensive label.

What millimetres per metre means

A specification in millimetres per metre expresses a small inclination as a rise divided by a horizontal run. A slope of 1 millimetre per metre means a height change of 1 millimetre across 1000 millimetres of run. The ratio is 0.001.

For an angle theta, slope equals tan(theta). Therefore theta is arctan of rise divided by run. At small angles, the angle in radians is approximately equal to that ratio. This connects a practical workshop unit with ordinary trigonometry.

Do not confuse a tool’s maximum permissible indication error with a required construction slope. A marking of plus or minus 0.5 millimetre per metre can describe instrument performance; it does not tell a builder that every floor should slope by that amount.

Nor is it a promise about the completed installation. Contact errors, reference choice, movement during adjustment and the object’s own shape can add uncertainty. Instrument specifications describe the instrument under stated conditions, not every possible measurement made with it.

STABILA’s Type 80 AS-2 specification gives different normal and reversed-position accuracies for that particular model. This illustrates why orientation and stated conditions belong with the number; the figures should not be applied to unrelated levels.

Worked example: a small stated inclination

Suppose an illustrative instrument tolerance is plus or minus 0.5 millimetre per metre in the position being used. Across a 1.2-metre horizontal span, the corresponding height difference is 0.5 multiplied by 1.2, or 0.6 millimetre.

Expressed as an angle, the ratio is 0.5/1000 = 0.0005. The arctangent is approximately 0.0286 degree. A small angle can therefore matter when carried across a sufficiently long distance.

This calculation converts a stated inclination; it does not certify the measurement setup or the part. A small chip under one end of the level can create an error comparable to the nominal tolerance. The arithmetic is exact only to the extent that its assumptions and rounded inputs are appropriate.

A useful report might state the tool, orientation, contact region and observed result. It should not announce that the entire object is accurate to 0.6 millimetre merely because that number emerged from multiplying a specification by length.

Worked example: percent slope is not degrees

A 2% slope means rise divided by run equals 0.02. Across one metre, that is 20 millimetres of rise. The angle is arctan(0.02), approximately 1.15 degrees, not two degrees.

Conversely, a two-degree inclination has a slope of tan(2 degrees), about 0.0349 or 3.49%. The difference is substantial enough to matter when someone confuses a percentage mark with an angle mark.

These examples illustrate units only. They are not a recommendation for a drain, accessible path, roof or floor. Required slopes depend on the application and applicable design requirements. The measurement tool helps check the chosen requirement; it does not choose that requirement for you.

Some specialised levels include marked pitch positions. Use those positions only according to their documented meaning. An ordinary pair of centring rings cannot be converted into a percent slope by assuming every manufacturer chose the same curvature and spacing.

Worked example: bubble travel and curvature

For a simplified small-angle model, let the vial’s relevant internal curvature have radius R. Bubble travel along the curved surface is approximately R multiplied by the tilt angle in radians. This is the familiar arc-length relationship applied locally.

Assume an illustrative radius of two metres and a tilt of 0.001 radian. The predicted bubble travel is 0.002 metre, or two millimetres. The corresponding small slope is about one millimetre per metre.

The example explains why a very gentle internal curvature can make tiny tilts visible. It does not mean the physical vial is two metres long. A short segment can be part of a much larger-radius curve, just as a small arc can belong to a very large circle.

Real vials have particular geometry, bubble size and manufacturing tolerances, so use the stated sensitivity for numerical work. The model teaches the relationship without pretending to reverse-engineer an unknown tool from its appearance.

A reliable horizontal reading begins at the contact

Identify the intended measuring face and remove loose contamination from both the face and the workpiece. Place the tool so it represents the line you want to check. Allow the bubble to settle before judging its position.

Keep the tool in stable contact rather than pressing one end down selectively. If the workpiece is flexible, hand force can change its shape while the measurement is being taken. The bubble then describes the loaded condition, not necessarily the object’s undisturbed position.

Read the bubble against its reference marks from a suitable viewing position. A sideways view can make relative positions harder to judge. Good lighting helps, but strong reflections or a scratched vial can obscure the edges.

For a plane such as a tabletop, check two independent directions and consider additional locations. A centred bubble along the front edge cannot establish that the back edge has no twist. Keep the spatial scope of the conclusion matched to the locations actually checked.

For the structural question of why a shelf may sag while its supports remain level, use How shelves carry loads. Orientation and bending are separate properties.

Reversal checks reveal a reference error

A reversal check compares the instrument with itself after its direction is changed on the same stable reference. The purpose is to distinguish an inclination of the support from an offset in the instrument’s indicated zero.

For a simple centred check, use a stable surface or two secure supports and establish a position in which the bubble is centred. Mark the tool’s footprint. Turn the level end for end by 180 degrees in the horizontal plane, keeping the same measuring face down and returning it to the same contact locations.

If the bubble no longer centres, something in the tool-reference relationship or contact setup needs investigation. Check for dirt, a changed support position and a damaged measuring face before blaming only the vial. Repeat carefully rather than adjusting a vial from one inconsistent attempt.

Johnson Level’s accuracy-check guidance uses reversal on a marked surface. This is a practical check, not a replacement for formal calibration when a demanding measurement requires it.

Do not confuse end-for-end reversal with turning the tool upside down onto a different face. An inverted-position measurement can have a different specification or may not be supported by the instrument. Maintain the intended contact face for the simple check described here.

Understanding reversal without a sign mistake

Suppose a numerical inclinometer reports signed readings in its own body direction. Let s represent the actual support slope and b represent an instrument offset. In one orientation, a simplified model gives r1 = s + b. After end-for-end reversal, the support slope changes sign in the tool’s coordinates: r2 = -s + b.

Adding the readings gives r1 + r2 = 2b, so the estimated offset is half their sum. Subtracting gives r1 – r2 = 2s, so the estimated slope is half their difference. The algebra separates the two contributions under the model’s assumptions.

For an illustrative example, r1 is positive 0.8 millimetre per metre and r2 is negative 0.4 millimetre per metre. The estimated offset is positive 0.2, and the estimated support slope is positive 0.6 in the first direction’s convention.

An ordinary bubble vial may not provide calibrated numerical divisions. Do not invent these readings from unlabeled rings. The calculation explains why reversal is informative; the centred practical check avoids needing to assign unsupported numbers to bubble displacement.

If a sloping surface is used without first centring, remember that the physical uphill end and the instrument’s labelled ends exchange roles. A bubble that stays toward the same high side of the room is not necessarily staying on the same side of the tool. Clear reference directions prevent a false diagnosis.

Straightness and local bumps can fool the conclusion

A level placed on two raised spots adopts the line between those supports. It may bridge a hollow in the middle. The bubble can be perfectly centred even though the workpiece is not flat underneath the entire tool.

Likewise, a single grain of grit under one end changes the tool’s inclination. For a rough estimate, a 0.5-millimetre lift over a 500-millimetre support separation produces a slope of 0.001, or one millimetre per metre. A tiny visible contaminant can therefore matter more than expected.

The support separation is the relevant length in that calculation, not automatically the tool’s total printed length. If the tool contacts only two narrow regions closer together, the same height difference creates a larger angular change.

Check contact directly where practical. Look for rocking, gaps and debris without putting yourself near moving equipment or unstable structures. A spirit level is not a substitute for a straightness survey, and neither measurement should be claimed from the other’s result alone.

Circular vials, torpedo levels and long levels

A circular or bull’s-eye vial uses a curved surface that allows the bubble to move in two directions. Centring it indicates that its mounting plane is in the intended orientation. This is useful for an instrument base or platform whose tilt is not confined to one axis.

A line vial offers a more directional indication and may make small changes along that axis easier to inspect. Circular and line vials therefore trade different kinds of information; neither is universally more sophisticated.

Short torpedo levels are convenient in confined spaces and on small components. Long levels sample broader spans and can bridge irregularities. The right length follows the feature and access, not a general rule that the longest tool is always most accurate.

Magnetic bases free the user’s hands on suitable ferromagnetic surfaces. They can also attract metal filings. Before trusting a magnetic level, inspect the contact face carefully: debris held by the magnet becomes part of the measurement geometry.

Digital levels and phone displays

An electronic level can infer inclination from a sensor responsive to gravity and display an angle or slope. This makes numerical comparisons convenient, particularly away from the horizontal or vertical references of a simple vial.

However, acceleration is not always gravity alone. Movement, vibration and handling can disturb the inferred angle. The device’s case and contact surface also matter. A phone with a protruding camera or a flexible cover may rest at a different angle from the surface you intended to assess.

Before comparing digital results, check units, zero mode and the reference face. A relative-zero setting can make a tilted surface read zero deliberately. That is useful for comparison but misleading when another user assumes the display means horizontal relative to gravity.

A laser level solves a related but different task by projecting a reference line or plane. It still requires appropriate setup and interpretation. A bright line on a wall is not proof that every object touching the line is straight or structurally sound. Follow the specific instrument’s instructions and laser safety guidance.

Safe learning activities that isolate one idea

Compare two directions on a board

Place a rigid board securely on a table and keep hands away from pinch points. Read a level along one direction, then across it. Adjust nothing while recording the first pair. Describe whether the board’s orientation differs between directions.

Repeat at another location. If the readings vary, consider whether the board is warped, the table has irregular supports or the tool was positioned differently. The objective is to identify possible causes, not to force every observation into the claim that the tool is broken.

Use a known small height change

On a stable tabletop arrangement, a thin card under one end of a supported rigid strip can create a visible change. Predict which way the bubble will move before inserting the card. The result checks the direction of response without requiring a numerical calibration.

Measure the support separation and card thickness only if the available measuring tools are suitable. Several stacked cards do not necessarily equal an exact multiple of one loosely measured card because compression and thickness variation matter. State the approximation rather than reporting unnecessary decimal places.

Test a reference reversal

Mark the contact locations, centre the bubble using a stable arrangement and reverse the tool end for end. Keep the contact face and supports unchanged. Record agreement or disagreement rather than attempting to repair the vial mounting.

This activity teaches controlled comparison. The level is both the instrument being checked and part of the experimental setup, so contact stability matters. A repeatable disagreement is evidence to investigate, while one rushed mismatch is not a complete diagnosis.

Interpreting a specification without overclaiming

A stated accuracy belongs to a model, measuring position and set of conditions. It should be read alongside length, intended measuring faces and temperature range. Similar-looking tools may have different capabilities even if their bubbles appear equally easy to centre.

A tolerance on the instrument does not include every error in an installation. The person may read from an angle, the object may move after the tool is removed, or fasteners may distort the part during tightening. The final result needs its own check.

For a demanding measurement, ask what uncertainty could change the decision. If the allowable difference is much smaller than the tool and method can support, use a more suitable procedure. Repeating an unsuitable measurement does not automatically make it adequate.

For an ordinary household question, avoid unnecessary precision theatre. It may be enough to establish that a surface has a noticeable tilt, then adjust and verify according to the object’s instructions. The appropriate standard is fitness for the task, not the maximum number of digits available elsewhere.

Troubleshooting by symptom

The bubble changes when the tool is pressed

The tool may be rocking on uneven supports, the workpiece may be bending or the body may be flexing. First inspect contact and remove selective hand pressure. A reading that depends on how hard one end is pressed is not yet a stable description of the undisturbed surface.

The bubble sticks or splits

Inspect the vial for damage and follow the maker’s instructions. Do not strike the tool to force a desired reading or attempt to refill the chamber. The vial’s fluid volume, surface condition and mounting are parts of the measuring system.

Reversal gives different results

Confirm that the same face rests on the same locations and that nothing shifted. Clean the contact surfaces and repeat. Persistent disagreement can indicate an instrument problem, but formal adjustment or replacement should follow suitable verification rather than guesswork.

Horizontal looks correct but vertical does not

Different vials and contact faces may be involved. Check that the intended vertical vial is being used and that the object is assessed in the relevant directions. Agreement in one orientation does not prove accuracy in every other measuring position.

A long level and a short level disagree

They may be sampling different support regions. A long tool can bridge a depression that a short tool rests inside. Compare them on a suitable shared reference before deciding that their different readings are solely instrument error.

Care that preserves the reference

Keep the measuring faces clean and store the instrument where its body and vials are protected from impact. Wipe contamination using methods compatible with the materials. Avoid strong solvents or excessive heat that may damage transparent components, markings or mounting arrangements.

Inspect after a significant fall or unexpected reading. A tool that survives without a visible crack may still have altered alignment. Verification matters more than the reassuring appearance of an intact case.

Label an instrument that is unreliable so it is not returned casually to important work. The aim is not to discard every scratched tool, but to distinguish cosmetic wear from changes that affect the measurement reference.

Checking the result after the work changes

A reading taken while an object is loosely supported may not describe its final position. Tightening a connection, loading a shelf or moving a support can change orientation. Measure the relevant finished condition rather than assuming that an earlier centred bubble remains valid indefinitely.

This is particularly important when a flexible object rests on several supports. Adjusting one support can redistribute load and change contact elsewhere. A local improvement is not automatically an improvement of the whole surface. Recheck the required directions and locations after the arrangement has settled.

For a hypothetical display shelf, record the empty condition separately from the loaded condition. A change under load may reveal bending or support movement. Do not disguise the distinction by describing both observations simply as level. The load condition is part of the result.

The instrument cannot decide which condition is acceptable. That comes from the object’s design and intended use. It can help reveal that conditions differ, giving a more specific question to the person responsible for the installation.

A record that preserves direction and place

A note saying left side high becomes ambiguous when the observer walks around the object. Use a fixed reference such as the marked end A, the door side or a sketch. Mark the tool’s orientation on that sketch so a later reading can be compared without guessing which direction was called positive.

Record which face of the level contacted the object and whether the check was horizontal, vertical or relative to another surface. This is especially useful when a tool has multiple vials or a digital relative-zero function. Two people can otherwise record matching zeros while referring to different orientations.

For a simple bubble instrument, avoid inventing numerical precision. A description such as centred within the reference marks in both marked directions is more honest than assigning an angle that the vial does not resolve. A photograph can support the record, but perspective and reflections may still make it less definitive than direct observation.

When a result is close to an important boundary, state what was not controlled. Contact roughness, unstable supports or a recent impact to the tool can all limit confidence. A clear limitation lets the next person improve the method instead of repeating an unexplained number.

Why agreement between tools needs context

Two levels agreeing on one surface is useful, but not conclusive. Both could share a similar offset, or both might rest on the same pair of high spots. Agreement becomes stronger evidence when the contact geometry, reversal checks and intended measuring directions have also been examined.

Conversely, disagreement does not automatically mean that the more expensive or newer tool is right. Compare both on an appropriate stable reference and inspect each tool’s measuring face. A clean modest instrument may outperform a damaged premium one in a particular task.

The same principle connects spirit levels with other instruments: compare the conditions as carefully as the indications. Reliable measurement comes from a controlled relationship between tool and object, not from brand confidence or the visual neatness of a bubble.

Frequently asked questions

Why does the bubble move toward the higher end?

The bubble is less dense than the liquid and occupies the highest available region of the curved chamber. When the instrument tilts, that region shifts. The bubble therefore moves uphill rather than behaving like a heavy ball rolling downward.

Does a centred bubble prove the surface is flat?

No. It indicates the orientation of the tool on its actual supports. A hollow or bump may remain beneath it. Flatness and straightness need appropriate contact comparisons, while a level checks inclination relative to gravity.

Must every level have a machined metal face?

No. Different constructions can provide a suitable reference surface. What matters is the instrument’s verified relationship between measuring face and vial, its specification and its condition. Appearance alone is not a calibration certificate.

Can I read a slope from the bubble rings?

Only when their numerical meaning is specified. Many rings simply guide centring. Specialised pitch vials may provide explicit slope marks, but those values cannot be transferred to an unrelated ordinary level.

Is two percent the same as two degrees?

No. Two percent is a rise-to-run ratio of 0.02, corresponding to about 1.15 degrees. Two degrees corresponds to about 3.49 percent. Always identify whether the requirement uses an angle, a percentage or millimetres per metre.

Is a longer level always more accurate?

No. Length changes the span being sampled. A longer tool can be useful for a broad line but may bridge local defects. Accuracy still depends on construction, specification, contact and condition.

Can I turn the level upside down?

Only use measuring positions supported by that instrument. Some models have different tolerances in normal and inverted use. Turning end for end on the same face for a reversal check is a different operation from measuring on the opposite face.

What does a reversal check establish?

It helps reveal disagreement between the tool’s indication and a stable reference arrangement. It is a useful check, not a universal calibration certificate. Keep the same supports and intended measuring face, and investigate contact errors before altering the instrument.

The bigger picture: gravity becomes a practical reference

A spirit level makes an invisible reference visible through a simple density contrast and controlled curvature. The bubble is only the final indicator. The complete chain includes the workpiece, contact face, body, vial mounting, liquid and observer.

Understanding that chain replaces vague conclusions with precise ones. The front edge is level is not the same as the whole surface is flat. The bubble returned after reversal is not the same as every measuring position is certified. A correct number or centred bubble must carry its conditions with it.

The most useful habit is to name the geometric question before reaching for the tool. Once level, plumb, straight, flat and square are separated, the instrument becomes easier to use and its limitations become easier to respect.

Useful routes

Use calipers and dimensional measurement for contact-based length readings and shelves and bending for the difference between inclination and deformation. Manufacturer sources linked above explain particular vial and instrument designs. All numerical cases here are illustrative calculations, not installation requirements or calibration certificates.