SMS character limits explained

An SMS has 140 bytes (octets, in the standards) for its content. The default character set, GSM-7, uses 7 bits per character, and 160 characters of 7 bits fill those 140 bytes exactly. 3GPP TS 23.038, the specification for SMS alphabets, says a message in the GSM 7-bit default alphabet can consist of up to 160 characters.

The same specification says the GSM 7-bit default alphabet has to be supported by every phone and service centre offering SMS, so it's the safe default for plain English. It covers A to Z in upper and lower case, the digits, everyday punctuation and a set of symbols and accented letters, including £, é, è, ü, ñ and à.

More than people expect, with a few gaps that catch them out. Alongside A to Z, the digits and everyday punctuation (including # % & * + = < > and round brackets), GSM-7 has £, $, ¥ and @, the Spanish ¡ and ¿, the § sign, and a run of accented and Nordic letters: à, è, é, ì, ò, ù, É, Ç, Ä, Ö, Ñ, Ü, ä, ö, ñ, ü, Æ, æ, Ø, ø, Å, å and ß. It even has ten Greek capitals, Δ, Φ, Γ, Λ, Ω, Π, Ψ, Σ, Θ and Ξ.

Texts tend to go wrong in the gaps. There's no á, í, ó or ú, no letters with a circumflex, no ë and no lower-case ç, so a customer called Seán or Zoë turns their own text into Unicode. That matters with merge fields: each person's text is counted after their name goes in, so one contact with an accent outside the set can take an extra part while everyone else's doesn't.

If a text contains even one character that has no GSM-7 form, the whole message is sent in UCS-2, the Unicode encoding SMS uses, at 16 bits a character. The space is still 140 bytes, so the limit drops to 70 characters. TS 23.038 says a UCS2 message 'can consist of up to 140 octets, i.e. up to 70 UCS2 characters'.

The usual culprits are characters you didn't notice you'd typed:

GSM-7 has an extension table for a few symbols that didn't fit in the main set. To send one, the message carries an escape code first and then the symbol, so each one takes two of your 160 places. TS 23.038 puts it as the number of displayable characters reducing by one for every time the extension table is used. These are the characters it covers:

A text of 159 characters that includes one euro sign takes 160 places, so it still fits in one part. Add one more character of any kind and it splits into two. The euro sign is still GSM-7, though, so the limit per part stays at 160 or 153; it never drops to 70 the way a curly quote makes it.

Merge fields affect the count too. In CaptiFi Connect, {first_name} is replaced by each person's name before the count is made, so a long name can push one person's text into a second part while everyone else's fits in one. Leave a few characters spare, and check the longest names on your list.

When a message runs past one part, it goes as several, each carrying a small header that tells the phone how to put them back together in order. TS 23.040 sets out what that header costs: a split GSM-7 text holds 153 characters per part (160 minus 7), and a split Unicode text holds 67 characters per part (140 minus 6 bytes, divided by 2). The phone shows the parts as one message.

Two more details from the same specification matter when you count. A two-place extension character is never split across two parts, and neither is a Unicode character, so now and then a part holds one less than the limit. CaptiFi Connect's counter follows those rules, so the number of parts it shows you is the number you're charged for.

TS 23.040 says a service centre handles each part of a split message like any other short message, and texts are billed the same way, by the part. In CaptiFi Connect, one credit is one part to one UK mobile. A 200-character GSM-7 text is two parts, so two credits per person; send it to 300 people and that is 600 credits.

Unicode can push costs up without anyone noticing. A 140-character message in plain GSM-7 is one part. Paste in a single curly apostrophe and it becomes 140 Unicode characters, which needs three parts (67, 67 and 6). Sent to 1,000 people, the same message goes from 1,000 credits to 3,000.

The longest text CaptiFi Connect will send is 10 parts: up to 1,530 GSM-7 characters, or 670 in Unicode. Anything longer is refused before sending, and you aren't charged for it.