Five Spec Gaps That Let Inconsistent Slate Through — and the One Line That Closes Each

The sixty-second version: when a slate delivery arrives with visible colour bands or tiles that vary 3 mm in thickness, the stone isn't usually bad and the supplier isn't usually cheating. The problem is almost always a gap in the specification — a dimension of the order that was left to the factory's judgement instead of nailed down in writing. There are five gaps that let inconsistency through: colour band, thickness tolerance, batch control, sample protocol, and arrival verification. Each one has a single sentence that closes it — a line you can copy into your purchase order today. Five gaps, five lines, and the roof you imagined is the roof that arrives.

Part of the Roofing Slate Guide's Buying & Importing section — the prevention half of the consistency story whose installation half lives in Sorting and Mixing Slate Before Installation, and whose factory floor half lives in How Roofing Slate Is Graded and Sorted.

I write specifications for a living. Not marketing copy — the kind of document that sits between a buyer's idea and a factory's production line, and decides whether the roof that arrives matches the roof that was imagined. I've been doing this for slate orders for eleven years, and if I had to name the single most common complaint that crosses my desk, it's this: "The slate arrived and it doesn't look like the sample." Or: "Half the tiles are 7 mm and the other half are 9." Or the quiet, miserable version: "The roof is finished and one slope is darker than the other."

Here's what I've learned from investigating every one of those complaints: in about eight cases out of ten, the stone was fine. The supplier didn't swap the batch. The factory didn't deliberately mix grades. What happened was that the specification left a door open — a dimension of the order that the buyer assumed would be handled, and the factory handled it the way they always do, which wasn't what the buyer had in mind. The inconsistency was built into the order before the first tile was split.

This page is about closing those doors. Five of them, in the order they matter. Each gap has a name, a picture of what goes wrong when it's left open, and the one line of specification language that shuts it. You don't need a fifty-page spec sheet. You need five sentences — and the discipline to put them on paper before the deposit clears.

Two roofing slate tiles from the same delivery showing visible colour difference on a white background

Gap 1: The Colour Band — When "Grey" Isn't a Colour, It's a Range

What goes wrong: the buyer says "dark grey," the factory produces dark grey — and the two sides mean different things by the word. The buyer is picturing the sample tile on their desk: a cool blue-grey, even and deep. The factory's "dark grey" runs from a warm charcoal to a greenish slate-blue, because the bed they're cutting that week leans that way. Both parties are acting in good faith. Neither is lying. The delivery arrives, the buyer opens the crate, and the tiles look like they came from a different quarry than the sample.

Why it happens: "grey" is not a colour in the slate trade. It's a family. A single quarry face shifts in tone as you cut through it — one week's output is darker, the next is slightly bluer, and there is no way to make natural stone match a single Pantone chip. What you can do is agree a range — a band of acceptable shades — and tie the sample to that band.

The colours guide explains how natural colour forms and why variation is normal. Here, the practical question is: how do you put "a range" on paper so the factory knows what you mean?

The line that closes the gap:

Colour: grey, matching the approved reference sample [sample ID / batch number]. Acceptable range: not lighter than [shade code] and not darker than [shade code]. Supplier to confirm in writing that the production batch falls within this range before production begins.

Two shade codes, that's the whole trick. You hold the lightest acceptable tile and the darkest acceptable tile from your sample set, assign them codes (your supplier can help — our factory uses A/B/C shade bands), and the "range" stops being a feeling and becomes a check. If the factory's current bed falls outside the band, they tell you before they cut — not after the container sails.

Six roofing slate tiles arranged from light to dark, each labelled with a shade code

And one thing that catches experienced buyers too: photographs are not colour references. A phone camera compresses the dynamic range, shifts the white balance, and flattens the exact difference you're trying to capture. The reference is the physical tile — the one you approved at sample approval — not a JPEG of it. If you're ordering by photo alone, Gap 1 is already open.

Gap 2: The Thickness Tolerance — When "7–9 mm" Hides Two Piles

What goes wrong: the order says "7–9 mm." The factory packs 7–9 mm. The buyer opens the crate, measures ten tiles, and discovers that five are clustered at 7.1 mm and five are at 8.8 mm. The nominal band is correct. The distribution is a disaster — because the roof now has thin tiles next to thick tiles, and the difference is visible from the ground.

Why it happens: "7–9 mm" is a band, not a target. A well-sorted batch has its thickness centred on the middle of the band — most tiles at 8 mm, a few at 7 and 9. A poorly sorted batch, or a batch where two runs were combined to fill an order, has two peaks: one at the thin end, one at the thick end. Both are technically "7–9 mm." Only one looks right on a roof.

The thickness guide covers what each band means for your roof. The grading and sorting guide shows how the factory separates thickness tiers on the bench. Here, the question is: what do you write in the order to prevent the two-peak problem?

The line that closes the gap:

Thickness: 7–9 mm band. Thickness distribution to be centred on 8 mm ± 1 mm. No more than 15% of tiles to fall at the extreme ends of the band (below 7.3 mm or above 8.7 mm). Supplier to provide thickness distribution data on request.

That sentence does two things. It tells the factory you know the difference between a band and a distribution — which alone changes how they sort your batch. And it gives you a right: to ask for the numbers. A factory that sorts properly can produce a thickness histogram in ten minutes; one that can't, won't — and that silence is your signal.

A digital caliper clamped on a roofing slate tile showing the thickness measurement

And the field check you can run yourself, the day the crate opens: pull ten random tiles, measure each at the centre with a caliper, and write the numbers down. If they cluster around one value, you have a single batch. If they split into two groups, you have two batches packed as one — and a phone call to make. This is the same check our pre-shipment inspection runs before the container seals, because the two-peak problem is the easiest one to catch and the hardest one to fix after the roof is laid.

Gap 3: Batch Control — When Two Batches Become One Order

What goes wrong: the order is 1,300 m² of 500×250, 7–9 mm, grey. The factory has 800 m² ready from this week's production and 500 m² from last week's run. They pack both into the same container because the order is one number. The buyer opens the crates on site and discovers that one end of the roof is a slightly different grey than the other — because the two batches came from different parts of the quarry face, a week apart.

Why it happens: "one order" is not the same as "one batch." A batch is a production run from a single bed, split and sorted on the same day by the same crew. It's the unit of consistency — everything within a batch shares a colour and thickness profile. When a factory combines batches to fill an order, the variation between batches becomes the variation across your roof. The factory isn't being dishonest; they're filling the order. But they're filling it with two different populations of stone.

The line that closes the gap:

Batch: the order must be supplied from a single production batch. If the order quantity exceeds one batch's output, the supplier must notify the buyer in writing before production, specifying which crates belong to which batch. Crates from different batches must be labelled with distinct batch numbers and packed in separate, clearly marked groups.

The key word is notify before production. Once the container is sealed, it's too late to separate batches — they're mixed in the hold. If the factory tells you in advance, you can plan: install batch by batch, slope by slope, and use the sorting and mixing method at the transition. If they don't tell you, you discover it on the roof — and the only fix is a teardown.

Two roofing slate crate labels showing different batch numbers for the same order

Here's the number that makes this concrete. Our 20ft containers typically carry about 1,130–1,450 m² of 7–9 mm slate — that's one batch's worth, give or take. When an order sits inside that range, single-batch supply is realistic. When it's 2,000 m², you're almost certainly getting two batches — and the spec line above changes from a preference to a plan. The container capacity guide has the numbers for every thickness band.

Gap 4: The Sample Protocol — When the Sample Doesn't Bind the Delivery

What goes wrong: the buyer approves a sample. The factory produces the order. The delivery doesn't match the sample. The buyer complains. The factory says: "The sample was from a previous batch — this batch is from a different part of the quarry." And they're not wrong, because nothing in the order tied the sample to the production run.

Why it happens: a sample without a batch number is a photograph, not a contract. It shows what the stone looked like — it doesn't bind the factory to produce what matches it. The fix is to make the sample part of the specification: give it an ID, tie it to a batch, and require the production run to fall within its colour and thickness band.

The line that closes the gap:

Sample: the approved reference sample [sample ID] is part of this specification. The supplier must confirm in writing that the production batch falls within the colour and thickness range of the reference sample. A retained copy of the reference sample, signed and dated by both parties, is binding on both sides. If the production batch deviates from the sample, the supplier must notify the buyer before production begins.

That sentence turns a sample into a sealed reference — the same concept our sample approval guide calls a "signed and dated physical tile." Two copies exist: one with you, one with the factory. When the delivery arrives, you compare a tile from the crate to the sealed sample. If it's within the band, the order is good. If it's outside, you have a document that says the factory agreed to the band — and that document is your conversation starter, not a argument about memory.

A sealed roofing slate sample in a clear bag with a handwritten label showing colour, thickness, batch and signature

Gap 5: Arrival Verification — When Nobody Checks Until the Roof Is Half Done

What goes wrong: the crates arrive, the roofer opens them on the scaffold, starts laying, and three courses in someone looks down and says "these don't look right." By then, two crates are on the roof and the rest are in the yard. The supplier is 9,000 km away. The evidence — the crate labels, the batch numbers, the packing list — is half-destroyed by the installation process.

Why it happens: because the arrival check was treated as a formality, not a gate. The inspection checklist exists for exactly this moment — the thirty minutes between the truck backing up and the first crate being opened. But if nobody runs it, the first inspection is the roof itself, and by then the evidence is in the skip.

The line that closes the gap:

Arrival check: the buyer (or their representative) must inspect the delivery against the packing list, the reference sample and the thickness specification before any crate is opened for installation. The inspection must be documented with dated photographs of crate labels, batch numbers and a random sample of five tiles per crate. No installation may proceed until the inspection is complete.

Five tiles per crate, a phone camera, ten minutes. That's the whole check. It catches the two-peak thickness problem before the tiles are on the roof. It catches the colour-band drift before the roofer has mixed two batches. And the photographs are your evidence if you need to have the conversation with the supplier — because "the tiles on the roof look different" is a complaint, and "crate 3, batch 047, tile 2: 8.9 mm, colour code C, outside agreed band B" is a claim.

The breakage allowance page explains the claim window — 3 to 7 days from delivery — and the same clock applies here. The longer the tiles sit before checking, the harder it is to prove the inconsistency came from the factory and not from site handling. Check early, photograph everything, and the five gaps stay closed.

The Five Lines on One Page: Your Spec Sheet

Here's the whole guide, compressed into five sentences you can paste into a purchase order. Print this. Attach it to the order. It's not a legal document — it's a shared understanding on paper, and that's what prevents inconsistency:

A printed specification sheet pinned to a factory wall with five lines highlighted
GapWhat it preventsKey phrase
1. Colour bandDelivery looks different from the sample"Not lighter than [code], not darker than [code]"
2. Thickness toleranceTwo thickness peaks inside one band"Centred on 8 mm ± 1 mm"
3. Batch controlTwo batches mixed as one order"Single batch, or notify before production"
4. Sample protocolSample doesn't bind the delivery"Reference sample is part of this specification"
5. Arrival verificationNobody checks until the roof is half done"Inspect before any crate is opened"

One last thought, and it's the one I give every buyer on their first order: inconsistency is not a quality problem — it's a communication problem. The factory that produced your slate can sort colour to a band and thickness to a distribution. They just need to know what you expect, in language that doesn't leave room for assumption. Five sentences on paper, and the roof you imagined is the roof you get. Leave them out, and you're betting the look of the whole roof on the factory's default — which is usually good, but is never exactly yours.

Frequently Asked Questions

Why does my roofing slate delivery look different from the sample?

Usually because the sample wasn't tied to the production batch. A sample without a batch number shows what the stone looked like, but doesn't bind the factory to produce within its range. Add a colour band and a batch reference to your purchase order — two shade codes and the sample ID — and the factory confirms in writing before they cut.

How much colour variation is normal in roofing slate?

Within a single batch, a spread of two to three shade steps is normal and expected — it's what gives a natural slate roof its character. What's not normal is a delivery that splits into two visibly different colour groups, which usually means two batches were packed as one. The fix is a batch-control clause in the spec, not a different supplier.

What thickness tolerance should I specify for roofing slate?

State the band (e.g. 7–9 mm) and the distribution: centred on the middle value, with no more than 15% at the extreme ends. This prevents the two-peak problem where half the tiles are at 7 mm and half at 9 mm, both technically inside the band but visibly different on the roof. Ask for thickness distribution data if you want to verify it.

Can I prevent colour mismatch between two batches of slate?

Yes, but only if you know the two batches exist. Require the supplier to notify you in writing before production if the order spans more than one batch, and to label the crates by batch number. Then install batch by batch, slope by slope, and use the three-pile sorting method at the transition to blend the boundary.

Should I check slate colour and thickness when the delivery arrives?

Yes — before any crate is opened for installation. Photograph the crate labels and batch numbers, pull five random tiles per crate, measure thickness with a caliper, and compare the colour to your sealed reference sample. The check takes ten minutes and gives you dated evidence if the delivery falls outside the agreed band. Waiting until the roof is half laid turns a fixable issue into a dispute.

What is a slate specification sheet and do I need one?

It's a one-page document that lists the five binding lines: colour band, thickness tolerance, batch control, sample protocol and arrival verification. You don't need a fifty-page spec — five sentences attached to the purchase order are enough to close the gaps that let inconsistency through. The line that matters most is the one tying the sample to the production batch.

Related Reading

Get the Five Lines on Your Order — Before the Deposit Clears

Every quotation we send carries the colour band, the thickness distribution, the batch number and the sealed sample reference — so the roof you imagined is the roof that arrives. ask for a quotation with the five spec lines · browse the roofing slate range · view our test results

Scroll to Top