Over the last two posts, I worked through five legal doublets I use in almost every contract. Part 1 covered the history and two pairs that are pure habit: terms and conditions, and null and void. Part 2 covered three that carry real weight: covenant and agree, indemnify and hold harmless, and successors and assigns.
This last part asks a different question. Does any of it depend on having a written contract at all? Then it ends with the case this series has been building toward.
A contract doesn't need paper
Everything so far has assumed a written contract. None of it actually depends on one.
An oral agreement is just as much a contract as a written one. The exceptions are narrow: Ontario's Statute of Frauds still requires writing for a few kinds of dealings, such as guarantees and interests in land. And nothing about a condition precedent requires it to be written down or labelled.
One case, two kinds of condition
UBS Securities Canada Inc v Sands Brothers Canada Ltd, 2009 ONCA 328, shows how far this goes. It rewards a careful read.
The main deal. In November 2006, over a handful of short phone calls, UBS and Sands Canada agreed to trade 100,000 shares in the Montreal Stock Exchange's parent company at C$50 a share, closing January 3, 2007. Nothing was signed.
The Court of Appeal upheld the finding that this was a binding contract. The test is objective: would a reasonable bystander, looking at the whole course of dealing, conclude the parties meant to be bound? Here they had agreed on price, quantity and closing date. The unsigned share purchase agreement drafted later was just a description of how the existing deal would be carried out, not a condition of the deal itself.
The resale. Relying on that deal, UBS made a separate oral agreement to resell part of the shares to a third party, Greenfair, at an agreed price and quantity. Its closing was expressly conditional on the Sands purchase closing first. The Court treated that resale as binding (paras 27, 54, 75).
That's a real condition precedent, but a particular kind. UBS and Greenfair had a complete contract the moment they agreed on price and quantity. What waited on the outside event was performance, not the contract's existence.
The approval argument. Sands Canada argued the opposite kind of condition. It said the exchange's approval requirement for any share transfer meant no contract with UBS ever came into existence, so there was nothing to breach.
The Court disagreed. Applying Turney v Zhilka, [1959] SCR 578, it asked whether the parties had agreed that an uncertain future event would decide whether the contract existed at all. Exchange approval wasn't that. Citing Waddams on contracts, the Court treated it as a condition of performing an agreement that was already complete.
So one case gives you both kinds of condition precedent, side by side:
- A condition precedent to formation: no contract exists until the event happens. Argued here, and rejected.
- A condition precedent to performance: the contract binds now, and performance waits on the event. Agreed orally here, and upheld.
The takeaway. A deal with no writing at all can be as binding as one on letterhead, and a condition precedent of either kind doesn't need to be written down to work. What decides which kind you have is what the parties actually agreed to make conditional, never the label.
What this has to do with a 2026 termination case
Every case in this series tells some version of the same story: a court asking what the parties actually meant, not which word or label happened to get used. The most recent example is the one closest to home for anyone with an employment contract.
On August 6, 2026, the Court of Appeal released Baker v Van Dolder's Home Team Inc, 2026 ONCA 568. Lower courts had split on whether "at any time" language in a without-cause termination clause breaches the Employment Standards Act, 2000. The Court held that it doesn't, not by that wording alone. Drawing on the Supreme Court of Canada's decision in Earthco Soil Mixtures Inc v Pine Valley Enterprises Inc, 2024 SCC 20, it said courts "should avoid a legalistic 'magic words' approach whereby the use of certain words considered in isolation will necessarily determine the legal effect of a contractual provision" (para 31). The motion judge had gone wrong by "reading the words literally and in isolation" and, in the Court's words, "strained to find an ambiguity where none existed" (para 53). The real question is what the contract as a whole, read in context, actually communicates.
That doesn't mean employment contracts are read like any other deal. The Court restated that they aren't ordinary commercial contracts. Employees have far less bargaining power, they're most vulnerable when the job ends, and if a termination clause can reasonably be read more than one way, courts prefer the reading that benefits the employee (paras 28 and 29). What the Court rejected was manufacturing ambiguity. A clause isn't ambiguous just because someone can imagine a second meaning. There have to be two reasonable readings of the contract as a whole (para 32).
That's the modern version of the problem doublets were invented to solve. In 1066, the worry was whether a reader in either language would understand the deal. In 2026, the Court of Appeal asked the same question in different clothes: would a reasonable reader, in context, understand the obligation? The Supreme Court of Canada said something similar in Sattva Capital Corp v Creston Moly Corp, 2014 SCC 53, which moved Canadian courts toward reading a contract's words in light of the surrounding circumstances. Nine hundred and sixty years apart, clerks and courts keep landing on the same idea: a contract only works if the person reading it understands it.
Context cuts both ways, though. In October, in Ghazvini v Canadian Imperial Bank of Commerce, 2026 ONCA 766, the Court of Appeal showed where Baker stops. Reading a contract as a whole doesn't rescue a clause whose words fall below the statutory minimum. The difference between the two cases came down to whether the "for cause" clause expressly preserved the employee's minimum entitlements. Words still matter. They just matter in context.
What this means for you
- Oral promises count. A phone call or a handshake can bind you, with only narrow exceptions.
- An unsigned draft doesn't mean there's no deal. If the essential terms are agreed, the paperwork may just be implementation.
- If a deal should depend on something, say how. Spell out whether the contract itself waits on the event, or only performance does.
- Context helps, but it can't fix a clause that breaks the rules. For employment contracts, courts read the whole document, but it still has to meet the statutory minimums.
What's next
If understanding is the real test, then who you're writing for changes what clear drafting looks like. The next posts in this series take one termination clause and write it three ways: in the overly legalistic style many templates still use, in plain language, and for a very specific reader who is decidedly not a corporate HR department. Same obligations in all three. Very different reading experiences.
