144 Tiles or Nothing: Checking a Mahjong Set Is Complete

Mahjong is the only corporate gift we handle where 143 out of 144 is a zero. A tote bag with a slightly crooked print is still a tote bag. A pen that writes is still a pen. A mahjong set missing one tile is not a slightly imperfect mahjong set — it is an object that cannot be used for its purpose, and the recipient discovers this in front of three guests, on the one evening of the year they wanted to use it.

Worse, you cannot fix it. There is no spare-parts channel for a single tile from a specific production run in a specific back colour with a specific engraving fill. This article is the composition of a complete set, the reasons sets arrive short, and two inspection methods — one that takes 90 seconds per set and one that lets you screen 200 sets without counting 28,800 tiles.

What a complete set actually contains

The standard set is 144 tiles, and it is 144 for a reason that is worth understanding rather than memorising. The suits are three families of nine ranks, four copies of each. The honours are four winds and three dragons, again four copies of each. Those two groups give 136 tiles, and the eight bonus tiles — four flowers and four seasons — bring the set to 144.

Group	What is in it	Copies	Tiles
Characters	Ranks one to nine	4 each	36
Bamboo	Ranks one to nine	4 each	36
Circles	Ranks one to nine	4 each	36
Winds	East, south, west, north	4 each	16
Dragons	Red, green, white	4 each	12
Flowers	Four distinct designs	1 each	4
Seasons	Four distinct designs	1 each	4
Total			144

Then there is everything that is not a tile, and this is where specifications go quiet. A set is normally expected to include dice, a dealer or prevailing-wind indicator, and a small number of blank spare tiles. Some sets add scoring chips or sticks; some add a cloth. None of these are guaranteed unless you list them. Write them into the specification by name and quantity, because “one mahjong set” on a purchase order is not a description of anything.

Why sets arrive short

Shortages are almost never theft or carelessness. They come from the way tiles are made and packed, and once you know the mechanisms you know where to look.

• The line counts trays, not tiles. Tiles are loaded into moulded trays and the trays go into the case. A packer verifies that the trays are full, at speed, by eye. One empty slot in the middle of a tray is the single most common defect in this product.
• Rejects are pulled and not replaced. A tile with a chipped corner or a short fill gets removed at inspection. If the replacement step is not enforced, the set continues down the line with 143 tiles.
• Blanks and white dragons get confused. Both are largely blank faces. A spare counted as a white dragon leaves the set short of a live tile and long on spares — and it passes a simple tile count.
• Flowers and seasons get mixed between variants. The eight bonus tiles differ between designs. A tray topped up from a neighbouring run can give you eight bonus tiles that do not form two complete groups of four.
• Accessories are packed separately. Dice and the indicator are often bagged and dropped in at the end, which makes them the easiest items in the whole product to miss.

Notice that three of the five failures survive a naive count to 144. That is why counting alone is not an inspection.

The wall test: a 90-second completeness check

The game itself gives you the fastest possible verification, and almost nobody uses it. At the start of a hand, the tiles are built into four walls of eighteen stacks, two tiles high. Four walls, eighteen stacks, two tiles: 144. The wall is a counting device.

Ninety seconds for the wall, another two or three minutes if you sort the groups. It is the check to run on the pre-production sample, on the first article, and on every set you sample at goods-in. Run it alongside the physical inspection that tells you what tile density and size actually do to a set and you have covered both halves of quality in under ten minutes.

Screening 200 sets without counting 28,800 tiles

At our minimum order of 200 sets, building 200 walls is not realistic. Nor is it necessary, because a mahjong set has a property most gifts do not: it has a precise, repeatable total weight. Every complete set from the same run weighs the same. A set missing one tile is lighter by the weight of one tile, and on a full-size set that difference is far larger than the resolution of an ordinary platform scale.

So the method is: weigh every set, count a sample. Establish the reference weight from several known-good sets that have passed the wall test, then put each boxed set on the scale as it comes off the pallet. Anything outside the tolerance band goes into a pile for a full wall test. In practice this is one person, one scale and a few seconds per set.

Step	Coverage	Catches	Misses
Establish reference weight	Several verified sets	Sets the baseline and the tolerance band	Nothing yet — this is calibration
Weigh-screen	Every set	Missing tiles, missing dice, missing indicator, missing trays	Wrong tile substituted for a right one
Wall test on a sample	A defined sample of the lot	Count errors	Composition errors, unless you sort
Group sort on the same sample	Same sample	Duplicates, wrong flowers or seasons, blanks passed off as dragons	Cosmetic defects
Branding fan check	Same sample	Logo drift across tiles	Nothing relevant at this stage

Be honest about the one thing weight cannot see. If a packer replaced a chipped tile with any other tile, the count and the weight are both correct and only the sort will find it. That is why the group sort stays in the plan even though it is the slowest step, and it is also why identical backs are what keeps a marked set playable matters here — a substituted tile from another run will often reveal itself as an odd back long before anyone notices the face.

How big should the sample be?

Do not invent a number. The sampling table used across this trade is the published standard for inspection by attributes, ISO 2859-1. Under general inspection level II, a lot of between 281 and 500 units falls in code letter H, which sets a sample of 50 units. For a 200-set order, that is 33 sets fully wall-tested and sorted, on top of the weigh-screen across all 200. Write the standard, the level and the acceptance criteria into your purchase order and the argument at goods-in is over before it starts.

What one missing tile actually costs

This is the part buyers underestimate. A single tile cannot be sourced after the fact in a way that is acceptable in a gift. It would need to match the mould, the back colour, the surface gloss, the engraving depth, the fill colour and any branding, from a run that has finished. Even where a factory can produce one, the lead time to make, decorate and ship a single tile is the same as the lead time for a batch, and the tile that arrives will be visibly newer than its 143 siblings.

So the remedy for this product is not spare parts. It is spare sets. Order a small buffer of complete sets above your headcount and treat them as the replacement stock. A defective set is swapped whole, the recipient never sees a mismatched tile, and the buffer doubles as cover for the last-minute additions to a guest list that happen on every corporate mahjong set gift programme. This is the opposite of the advice for a printed mug or a lanyard, where spares are cheap and interchangeable.

The goods-in checklist

Do this at goods-in and not at the venue. Once sets are stacked in a ballroom store room they will not be inspected, and the first person to open one will be the recipient. The way the sets are packed decides how practical that inspection is, which is part of the wider question of how the set is boxed and carried home.

Common questions

How many tiles should a set have?

144 for the standard set: 108 suit tiles, 28 honour tiles and 8 bonus tiles. Some sets supplied regionally carry additional tiles used in local variations. Confirm which you are ordering and keep it consistent across the whole delivery.

Can we buy replacement tiles later?

Realistically, no — not in a way that matches. Plan for whole-set replacement instead and hold a buffer of complete sets from the same run.

Do we have to open every box?

No. Weigh every box and open the sample plus any weight outliers. That gives you full coverage on the failure that matters most, at a fraction of the effort of opening 200 cases.

What is the minimum order and the lead time?

200 sets. Production is 20 to 30 working days after artwork approval depending on quantity, plus around 15 days shipping. Express at 7 working days is available on limited models and quantities. Add a few days at your end for inspection before distribution.