Marine Sortable Silt, Mean Grain Size and Bottom Currents

Marine sortable silt can tell us something about the strength of currents near the deep seafloor, but a larger average grain size is not a universal ocean-speed meter. Translating this proxy means recognising which particles are included, how they were measured, what physical sorting process selected them, and whether changing sediment supply or preservation could mimic the signal. A value of 30 micrometres can be an excellent observation without constituting a direct measurement of a current flowing at 30 centimetres per second.

The key idea is that noncohesive mineral particles of different sizes do not settle and resuspend in exactly the same way. Under suitable deep-sea depositional conditions, stronger flows preferentially deliver, retain or expose a relatively coarser silt population. Researchers use the mean size of a carefully defined terrigenous fraction to infer changes in near-bottom flow. The inference is powerful in well-selected settings, but it is conditional and needs sedimentological context. This guide shows the units, calculations, competing explanations, and independent checks.

What is sortable silt, exactly?

The conventional sortable-silt window is roughly 10 to 63 micrometres in the terrigenous, noncohesive fraction. A micrometre, abbreviated µm, is one millionth of a metre. The lower cutoff avoids much of the more cohesive behaviour associated with finer silt and clay, while the upper cutoff matches the common boundary between silt and sand. Different analytical methods and sediment classifications should be checked for exact thresholds before datasets are merged.

The term ‘sortable’ does not mean every grain in this range has been individually labelled by a machine as current transported. It identifies a physically motivated size class that responds to sorting under relevant conditions. The fraction is typically assessed after considering or removing biogenic components such as carbonate and opal when the method calls for the terrigenous signal. If the dataset instead includes abundant foraminiferal fragments, a large mean may not represent the same process as a measured lithogenic sortable-silt mean.

Two statistics are commonly informative. Mean sortable-silt size, often written with a bar over SS, describes the average size of material within the selected fraction. Percentage sortable silt, SS%, describes how much of a specified reference sediment fraction lies between the cutoffs. They have different units: the mean is in micrometres and the fraction is a percentage. A study must define the denominator of SS%, such as the fine terrigenous material, rather than assuming all reported percentages mean percent of whole wet sediment.

Why currents can change the grain-size distribution

Settling velocity generally depends on particle size, density, shape and fluid conditions. Fine cohesive grains can aggregate into flocs, while coarser noncohesive silt behaves differently during suspension, settling and resuspension. In the deep ocean, episodic strong near-bed flow and bottom storms can influence which grain sizes are deposited or removed. The preserved size spectrum thus contains information about boundary-layer processes that operated during sediment accumulation.

At some flow conditions selective deposition preferentially retains certain size classes, while stronger events can winnow fine particles and leave a coarser assemblage. These mechanisms are related but not identical. A change in grain size could arise because newly arriving particles were sorted during deposition, or because older fine particles were later removed. The resulting layer may still look reasonably undisturbed in a broad photograph, so detailed sedimentological examination helps distinguish mechanisms.

McCave and Hall’s reviews emphasise that sortable silt is grounded in particle-transport physics but affected by site conditions. It has often been applied to sediment drifts and other settings where local changes in bottom flow plausibly shape terrigenous grain sizes. The proxy does not reconstruct every current in the basin. It offers evidence about flow conditions influencing deposition near the sampled seafloor, at times and spatial scales set by the sedimentary archive.

Current speed and current direction are separate problems

A size distribution can reflect relative flow intensity without telling us which way the water moved. Sediment grains have no intrinsic arrow indicating their transport direction. An increase in average sortable-silt diameter should not be translated automatically into a stronger northward current. Determining direction requires other evidence, such as seabed bedforms, current-meter observations, regional circulation patterns, anisotropic sediment structures or complementary geophysical interpretation.

Deep-bottom current strength is also not the same as the speed of surface currents. A strong surface eddy can exist above sluggish bottom water, while a dense overflow may generate energetic flow at depth without a similar surface expression. A core’s sortable-silt signal should be discussed in the water-mass and depth context of the sampling site. This is particularly relevant when an author proposes that a local grain-size change represents a basin-scale circulation reorganisation.

The meaningful result may therefore be ‘relative increase in local depositional flow intensity’ rather than ‘the named ocean current accelerated by a specified number of centimetres per second’. The second statement requires local calibration and independent dynamical evidence. Readers should notice when a manuscript moves from a measured mean grain size to a named global current system, and ask what observations connect those two claims.

Sample preparation is part of the meaning

A marine mud sample contains mineral grains, biogenic carbonate, biogenic silica, organic matter and potentially aggregates or cement. Protocols for sortable-silt analysis can remove or separate particular components to obtain a target terrigenous distribution. The treatment must be suited to the local sediment and carefully controlled; chemically altering mineral grains or inadequately dispersing aggregates may shift the apparent grain-size curve. A size analysis is only interpretable when the material being sized is defined.

Pre-treatment can involve decisions about carbonate dissolution, organic removal and particle dispersion. Removing abundant shell fragments may make the terrigenous flow-sensitive fraction clearer, but harsh or inconsistent procedures can damage other constituents or cause unequal losses. A study should state whether percentages refer to the original bulk sediment or the residual material after pre-treatment. Otherwise, apparent shifts in fraction abundance may arise from changing amounts of material removed rather than real changes in terrigenous sorting.

Dry mass, sample depth, extraction efficiency and replicate variability should be documented. Near an event bed, a small subsample can be unrepresentative if coarse grains are unevenly distributed. When comparing cores processed in different laboratories, pay attention to pre-treatment and size-window definitions. The biggest risk is not that the calculations are difficult; it is that the two reported numbers may describe chemically and physically different populations of grains while using the same familiar name.

Measurement methods do not all report identical sizes

Settling-based instruments infer size from hydraulic behaviour, whereas laser diffraction measures light scattering interpreted through optical and particle models. These are not guaranteed to return identical numerical diameters for platy, elongate or aggregated natural particles. McCave and Hall discuss concerns about laser-diffraction methods for flow-related fine particle sizing, including the influence of shape. The chosen size method belongs in the figure caption or supporting data, not merely in an inaccessible laboratory note.

An equivalent spherical diameter is a convenient instrument representation. A plate-like clay grain and a near-spherical mineral grain can produce different optical or settling responses even if their dimensions overlap. If mineralogy or grain shape changes through the core, an apparent size trend may partly reflect measurement physics. Comparing samples by the same carefully controlled instrument is valuable, but it cannot eliminate all source and shape effects. Independent grain analysis may help where the environmental interpretation depends on subtle differences.

Before converting a grain-size number into a paleoflow claim, ask how the instrument’s size bins, detection range, dispersant, pump or settling assumptions were set. Check replicate spread, the reference standard and whether material above 63 µm was correctly excluded. A plot generated from a software default may include sand or fine clay in the average, even when the article calls the result ‘sortable-silt mean’. A valid proxy starts with the right arithmetic population.

Calculate a size-restricted mean correctly

Suppose a constructed dataset has three representative bins within the sortable window: 15 µm carrying 20 percent of sortable-silt material, 25 µm carrying 35 percent and 40 µm carrying 45 percent. The weighted mean is 15×0.20 + 25×0.35 + 40×0.45, equal to 29.75 µm. Those percentages sum to one because the calculation has been restricted and renormalised to the material inside the selected fraction. If the bins did not sum to one, divide by their total weight.

A common mistake is to mix whole-sample percentages with already normalised within-window percentages. If the three size bins constitute only 30 percent of the original sediment, their raw whole-sample weights may be 6, 10.5 and 13.5 percent. Computing 15×0.06 + 25×0.105 + 40×0.135 would give 8.925 µm, which is not the average diameter of particles in the window. The correct weighted mean divides that weighted sum by the included fraction 0.30, recovering 29.75 µm.

Another mistake is to average the labels of size classes without weighting by their abundance. For the constructed bins, the unweighted average of 15, 25 and 40 is about 26.67 µm, different from the correct 29.75. A logarithmic or geometric mean can be appropriate in some grain-size conventions, but it is a different statistic with a different definition. Use the exact metric reported by the source, including whether size was handled on a linear or logarithmic scale.

Percent sortable silt needs a named denominator

A reported SS% of 25 may mean that one quarter of a defined terrigenous fine fraction falls in the 10–63 µm window. It does not necessarily mean 25 percent of the entire dry sediment core sample, let alone one quarter of all the original particles before carbonate or opal treatment. If a study reports an SS% change, inspect the preparation method and percentage basis. A correct number with an unspecified denominator can still lead to a wrong physical conclusion.

Imagine two constructed samples. The first contains 20 grams of residual terrigenous fines and 5 grams of sortable silt, giving SS% = 25. The second contains 10 grams of terrigenous fines and 3 grams of sortable silt, giving SS% = 30. The second sample has a larger sortable proportion but less absolute sortable-silt mass. This is not paradoxical; the denominator has changed. It warns against interpreting a fraction increase as a flux increase without mass accumulation information.

Mean size and percentage may move together or apart. A rising mean with stable SS% can signal a shift within the window, while rising SS% with stable mean can signal a change in the amount of silt present without much change in its internal size distribution. In a real core, either pattern could also reflect changing sediment sources. Reading both statistics, along with the full distribution, is more informative than choosing the one curve that best fits a preconceived circulation narrative.

Choose a suitable depositional environment

The sortable-silt method has been applied especially to settings where terrigenous particle sorting plausibly tracks variations in bottom-water flow, such as certain sediment drifts. The sedimentary setting should be examined before drawing conclusions. A deposit dominated by landslide debris, ice-rafted stones or episodic downslope flows may not preserve a simple relationship between mean size and background bottom-current strength. The presence of silt-sized particles alone is not enough.

Look for evidence of continuous accumulation, a stable or interpretable sediment source, and a depositional regime in which resuspension and sorting are plausible. Seafloor topography matters because flow can accelerate over ridges, decelerate in sheltered basins and form complicated patterns around obstacles. Two cores a short distance apart can therefore record different local conditions. Treat this spatial sensitivity as a feature of the proxy, not as an excuse to ignore regional circulation.

Even at a promising site, a change in sediment supply can alter the starting grain-size population. A stronger dust source, a river plume or a new glacial sediment input may shift the mean without requiring a corresponding change in bottom speed. Sedimentological and geochemical provenance evidence can help. The correct method compares alternative transport explanations rather than assuming all terrigenous grain-size change is hydraulic sorting at the core location.

Relative changes are easier to defend than universal velocities

A rise from 25 to 30 µm in the mean of the sortable fraction can be meaningful in a continuous, consistently processed record. It may indicate relatively more energetic deposition or winnowing under the right conditions. It does not establish that the bottom current increased by five centimetres per second. The units cannot be converted one-to-one because micrometres describe particle size while centimetres per second describe water velocity. A calibration would have to link them through measured flow conditions and sediment dynamics.

A few studies have explored empirical relationships between sortable-silt mean and near-bottom speed, including normalized comparisons across locations. Those relationships do not prove a globally universal absolute velocity calibration. Local sediment availability, particle shape, currents and topography influence the response. Even a useful regional calibration should state its measurement method, spatial extent and environmental conditions. Applying its numerical slope to an unrelated core without validation may produce misleadingly precise palaeovelocities.

A careful figure caption might say that the sortable-silt mean increased by 5 µm, consistent with stronger local bottom-flow sorting if terrigenous source and measurement conditions remained comparable. If site-specific calibration and independent evidence support an estimated speed change, the report can add that estimate with uncertainty. Otherwise, the measured micrometre shift is still useful. Scientific confidence comes from matching the claim to what the observations constrain.

Distinguish selective deposition from winnowing

Selective deposition occurs when a flow supplies and deposits a mixture of particles in size-dependent ways. Winnowing describes removal, commonly of fine material, leaving a coarser residual assemblage. Both can coarsen a grain-size distribution, yet their geological effects differ. Selective deposition may alter which particles arrive, whereas winnowing can remove material previously deposited. The mean size alone need not reveal which process dominates.

One possible diagnostic is the relationship between mean sortable-silt size and the proportion of sortable silt in the reference fine fraction. If both increase, that may be compatible with some sorting regimes, but the trend is not a unique fingerprint. A coarse-skewed distribution from a supply change can produce similar statistics. Sedimentary structures, bulk density, mass accumulation and provenance markers are needed to evaluate the mechanism. The best interpretation tests a physical sediment transport story rather than assigning meaning to the sign of one correlation.

Winnowing can also disturb the chronological record. An erosional hiatus may make an interval appear thinner than expected from the age model; coarse lag material may contain reworked particles. If a core has an apparent current-speed increase at the same time as an age discontinuity, the proxy cannot simply be interpreted as a smooth record of continuous flow. The inferred history might instead contain events, gaps and changing sediment sources.

Ice-rafted and downslope material can imitate a current signal

Ice-rafted detritus arrives by a different transport pathway from ordinary near-bed current sorting. It can introduce coarse mineral grains and unusual mineral compositions into marine sediment. Even smaller fragments can change the distribution inside a nominal sortable-silt window. A glacial interval with a new source of fine rock flour can therefore alter mean size independently of bottom-current strength. The presence of ice-rafted material should trigger stratigraphic and provenance checks before attributing grain-size changes to deep circulation.

Gravity flows can also deliver mixtures that were sorted upslope rather than at the core site. A turbidite may contain graded bedding and a grain-size pattern generated during its emplacement. If such beds are included in a long-term sortable-silt curve, they can generate spikes that seem to record vigorous background currents. A site-specific protocol should identify event deposits, state whether they were analysed separately, and explain whether their inclusion is appropriate to the research question.

Where sediment comes from may change through time. A river can shift drainage or sediment calibre; atmospheric dust sources can evolve; glacial weathering can alter grain shape and mineralogy. A deep current may remain similar while the supplied particle population changes. Geochemical provenance, clay mineral assemblages and textural comparisons can help evaluate these alternatives. No single grain-size proxy can separate current intensity from arbitrary source change without additional information.

Core-top calibration and spatial gradients

Modern sediment can provide a calibration opportunity if near-bottom flow has been measured independently. Yet a modern core top may integrate decades or centuries, while a current meter samples a much shorter time interval. Even if measurements are accurate, their averaging windows differ. A short deployment during an unusual season or storm may not represent the long-term depositional regime recorded by sediment. The time-scale mismatch belongs in the interpretation.

Spatial sampling creates a related problem. Velocity varies across a contourite drift and around seabed obstacles. A sediment core’s deposition can integrate lateral transport over an area, while a current meter records conditions at one location and height above bottom. A regression can be useful when sampling geometry is carefully designed, but matching a single current-meter number to one core size statistic should not automatically be described as a calibration for all ocean basins.

Published empirical slopes and standardized sensitivities are best treated as evidence for the existence of a relationship in studied settings, not as conversion constants. When a reader encounters a claim that 1 µm always equals a certain velocity increase, the correct response is to inspect the calibration study’s original sediment type, measurement method and data range. A credible quantitative reconstruction must demonstrate that those conditions transfer to the target core.

The full grain-size distribution can reveal more than its mean

Two samples can have the same mean sortable-silt size while showing different distribution shapes. One may have a narrow peak around 30 µm; another may combine 15 µm and 45 µm grains into a broader mixture. Their weighted means could be similar even though the sediment sources or sorting mechanisms differ. Frequency curves, cumulative distributions, modal structure and sorting statistics help determine whether a single average adequately describes the sample.

A bimodal distribution can be evidence that two source or transport populations are present. It is not automatically proof of two specific currents, because mixing, aggregation, pre-treatment and measurement artefacts can also create complex curves. Compare repeated subsamples, microscopy and neighbouring intervals. If a mean-size curve changes abruptly while the modal components stay fixed, the apparent trend may represent changing mixture proportions rather than a continuous shift in the hydraulic sorting of one population.

Endmember modelling can provide a structured way to describe mixtures, but its numerical components are not automatically physical sediment sources. The choice of number of endmembers, normalisation, data quality and model constraints affects results. A sophisticated decomposition is useful only when its components can be interpreted against field evidence. Otherwise, it can transform an uncertain physical story into apparently precise coloured curves without resolving the central question of transport mechanism.

How to handle carbonate, opal and organic-rich sediment

A deep-sea sample containing abundant foraminiferal tests can show an apparent rise in coarse size if biogenic particles are included. The terrigenous sortable-silt convention seeks to isolate the hydrodynamically informative mineral fraction, so sample processing often accounts for carbonate and sometimes siliceous components. But removal is a laboratory transformation that must be reported. What remains depends on the chemical treatment and whether delicate or reactive mineral grains were altered.

Biogenic opal may include fragmented radiolarians or diatoms whose shapes and sinking behaviour differ from mineral grains. Organic aggregates can bind fine grains into larger apparent particles until dispersion breaks them apart. If the degree of aggregation varies across samples or analyses, apparent coarsening can reflect preparation. Laboratory repetition with controlled dispersion and method comparison can help identify this issue. It is not always enough to label every particle of a given diameter as interchangeable.

A meaningful mass basis is especially important where the proportion of biogenic material changes greatly. After removing carbonate from a sample, a sortable-silt percentage calculated on the residual fraction cannot be compared directly with a pre-removal whole-sediment percentage. Both can be correct for their own denominators. Keeping original dry mass, removed mass, residue mass and fraction weights in the data table makes the transformation auditable.

A worked comparison of three environmental hypotheses

Consider two hypothetical intervals in one well-sampled core. Interval A has mean sortable-silt size 25 µm and SS% equal to 20. Interval B has mean size 31 µm and SS% equal to 28, using the same particle window, denominator and instrument. The simplest descriptive result is coarsening within the target fraction and an increased proportion of it. That is an observation. The physical explanation still needs to be tested against the provenance and stratigraphic context.

Hypothesis one proposes stronger bottom-current sorting during B. This would be more plausible if the site is a suitable contourite setting, the source material remains comparable, and independent seabed or regional circulation indicators align. Hypothesis two proposes a new supply of coarser terrigenous particles from an upstream source. It would gain support from a change in mineralogical or isotopic provenance. Hypothesis three proposes episodic winnowing or erosion that removed fine particles, potentially associated with an age hiatus or lag deposit.

These explanations predict different accompanying observations. A provenance switch may shift elemental ratios or mineral populations. An erosional event may alter chronology or create sedimentary structures. A sustained circulation change may appear coherently across multiple suitable core sites. The proper analysis states what each hypothesis predicts and then looks for measurements that could discriminate among them. A proxy curve becomes powerful when it is used to test mechanisms, not when a single trend is declared self-explanatory.

Statistics for a changing sequence of core samples

A long marine core may contain hundreds of samples with varying spacing. A trend line should respect that spacing and the age model. Computing an ordinary mean across all samples can overweight an interval sampled densely, while a time-weighted comparison may answer a different question. When reporting epoch averages, define the included time interval and how uncertainty in age assignments was handled. A numerically smooth series can still have uneven information content.

Replicates can estimate analytical or subsampling variance, but they do not remove systematic instrument bias or changing particle shape. If measurement protocols change halfway through a core series, the method change may produce a step that looks like a climate event. A bridging set of samples measured by both methods can help quantify an offset. Do not correct one period with an arbitrary constant merely to align it with the other; document the empirical basis and remaining uncertainty.

Autocorrelation and mixing affect inference about significance. Adjacent sediment samples are not always independent environmental observations because bioturbation and sampling overlap can blend the same material across intervals. A long run of closely spaced values may therefore contain fewer effective independent observations than the plot suggests. Confidence in a sustained trend should account for that structure as well as measurement error and age-model uncertainty.

What to do when the mean and SS% disagree

A higher sortable-silt mean with a lower sortable-silt percentage is not necessarily an error. It could mean that a smaller quantity of particles within the selected window became coarser while material outside that window increased. A lower mean with a higher percentage could mean more silt was deposited but with a finer internal distribution. The two quantities describe different properties and should not be forced into a single qualitative category.

To diagnose such a pattern, return to the actual grain-size distribution. Identify changes below 10 µm, within 10–63 µm and above 63 µm. Check whether the whole-sample and target-fraction denominators were handled consistently. Then inspect sample composition and transport setting. An apparent contradiction often disappears when the reader recognises that one statistic concerns conditional average size while the other concerns abundance relative to a broader reference population.

Even when both statistics move in the expected direction, the environmental conclusion remains conditional. Agreement is useful evidence, but it cannot independently validate the assumption of an unchanging sediment source. Compare neighbouring cores, grain chemistry and relevant physical oceanographic observations. A good paper shows the full distribution or makes it available for independent inspection rather than requiring readers to infer every change from two summary lines.

Transfer problem: a correct weighted mean and a wrong one

A constructed sortable-silt distribution has sizes 12, 24 and 48 µm with weights 10, 50 and 40 percent within the selected window. The arithmetic mean weighted by particle fraction is 12×0.10 + 24×0.50 + 48×0.40 = 32.4 µm. If the analyst simply averages 12, 24 and 48, the result is 28 µm and is not the fraction-weighted mean. The distinction matters because coarse grains are more abundant in this example.

Now suppose those same bins represent respectively 2, 10 and 8 percent of the original whole sediment rather than 10, 50 and 40 percent of the window. Their raw weighted sum is 6.48 µm when calculated as fractions of the whole sample, but the included weight is only 0.20. Dividing 6.48 by 0.20 restores the conditional mean 32.4 µm. A reported mean that is less than the lower cutoff of its own defined size window is a strong clue that the denominator was mishandled.

A further check concerns the measurement scale. If an instrument reports logarithmic phi classes rather than micrometre class centres, directly averaging phi values and converting the average is not generally identical to a linear arithmetic size mean. The paper must specify its statistic. Two studies can use the phrase ‘mean grain size’ while calculating different central tendencies. Comparability requires aligning definitions, not merely conversion of axis labels.

Transfer problem: no justified speed conversion

A core’s sortable-silt mean rises from 26 to 31 µm. How much did the current speed increase? The defensible direct answer is that the selected fraction’s mean size rose by 5 µm. Without a suitable site-specific relationship, no unique increase in centimetres per second follows. The proxy may be consistent with stronger relative flow, but current speed also depends on sediment supply, source properties and local transport. The solution should say which missing evidence would allow a stronger quantitative claim.

What if a published study provides an empirical sensitivity? Before using it, confirm its measurement method, environmental range, definition of mean and whether it estimates absolute or normalised speed. A slope derived from one sediment drift is not a law for every continental slope. If the calibration applies, propagate uncertainty in both mean size and slope. If it does not, report a qualitative relative interpretation and show the empirical comparison separately without pretending it is a direct measurement.

The wider lesson is that conversion of scientific meaning differs from changing units. Multiplying metres by one hundred gives centimetres because the relation is a definition. Turning micrometres of sediment size into centimetres per second of water flow requires an empirical model with assumptions. Confusing these operations produces confident-looking numbers unsupported by physics. A good reader can explain this distinction even without specialised oceanographic software.

Independent questions and worked answers

Question one: what particle-size range is conventionally called sortable silt? Answer: approximately 10–63 µm for the relevant terrigenous noncohesive fraction. State the exact method used in a real study, because analytical conventions and preprocessing matter. Question two: what is the unit of mean size? It is a length such as µm, not a speed. Question three: what is the unit of SS%? It is a proportion or percentage whose reference denominator must be defined.

Question four: in two synthetic samples the mean size remains 28 µm but SS% rises from 18 to 29. Has current speed necessarily increased? No. The internal mean of the window has not changed, while its proportion relative to the defined reference has increased. That could reflect changes in supply, sorting or other factors. A defensible analysis inspects the full size curves and independent sediment composition rather than assigning one physical cause.

Question five: a laser diffraction dataset and a settling-based dataset differ by 2 µm for ostensibly matching samples. Which method is correct? The information supplied does not establish one as correct. They may respond differently to particle shape, dispersion or calibration. Compare reference materials, replicate samples, protocol and method-specific definitions. Avoid treating a method-dependent offset as an environmental change.

Question six: a coarse spike coincides with an ice-rafted debris peak and an age-model discontinuity. Should it be interpreted as an abrupt strengthening of deep currents? Not without further evidence. Alternative sediment supply and event deposition explanations are clearly plausible. Isolate the interval and investigate its sedimentological properties before including it in a background paleoflow reconstruction. The strongest answer names a test capable of falsifying the easy story.

A practical data table and reporting workflow

Retain sample identification, core depth interval, age and its uncertainty, dry mass, material removed during pretreatment, target grain-size population, bin boundaries, measurement instrument and its settings, replicate records and full distribution. Derived fields should include mean sortable-silt size, fraction percentage with explicit denominator and any other statistical measure such as sorting or mode. Keeping the full histogram makes later reanalysis possible when methods or models improve.

The methods paragraph should state why the depositional setting is suitable, how biogenic components were treated, how particles were dispersed, what equipment was used and how the size statistics were calculated. The discussion should distinguish source changes, sorting and winnowing. It should not convert grain size to speed unless a defensible calibration is established. A brief uncertainty account should include subsampling, instrument response, pretreatment and geological interpretation, not only counting variation in repeated runs.

The conclusion can then be concise: the target fraction became coarser, by a quantified amount, under consistent measurement procedures; the pattern is consistent with a specified change in near-bottom sorting if the source and depositional assumptions hold; additional evidence supports or limits that mechanism. This is stronger than simply announcing that a current accelerated. The reader knows what the instrument observed and what the researcher inferred.

Frequently asked questions

Does sortable silt show the direction of deep currents? Not alone. It more directly relates to sorting and local flow intensity under appropriate conditions, while direction needs independent evidence. Can I use all silt from 2 to 63 µm? That changes the conventional noncohesive target by including much finer cohesive material. A separate analysis can be legitimate, but it should not be called the same index without explanation. Is a higher mean always stronger flow? No; particle supply, winnowing, biogenic contamination and instrument effects can also change the result.

Should an author publish both mean size and SS%? That is often helpful because they describe complementary aspects of the distribution. Should a deep ocean flow reconstruction use only grain size? No; independent provenance, stratigraphy, hydrography and geophysical evidence make the interpretation more robust. Can two laboratories be compared? Yes, after confirming size boundaries, pretreatment and measurement methods and considering systematic offsets.

What is the central translation skill? Know the difference between a micrometre-scale grain measurement and a current-speed inference. Calculate the restricted mean using the correct weights and denominator, understand the physical sorting mechanism, and test rival explanations against independent observations. That discipline makes a sortable-silt record intelligible without turning a useful proxy into a fictitious universal speedometer.

Editorial note: this original educational explanation synthesises the cited literature. Every numerical exercise is a constructed example rather than a measurement from an ocean core. AI assistance was used in drafting; readers should consult primary studies and real site data for field application.