Starc's Six at Newlands: The Truth About 6/72, and the Lie Inside the Five-For Table

In 2026 at Perth, Mitchell Starc bowled six South Africans out for 154. On Tuesday at Newlands he bowled six South Africans out for 72. Same bowler, same number, eighty-two runs of distance between the two spells. The scorecard hands both of them the same letter: six.

My argument is that the 6/154 was the better spell. This isn't nostalgia. Taking six on a WACA trampoline at 154 was the hardest job in cricket that day. The 6/72 at Newlands is a bill that South Africa's batters wrote themselves, and Starc merely collected it. The problem is that a scorecard does not distinguish collecting from earning. It just counts.

The biggest lie of the day wasn't the six wickets. It was 203/1.

More than two decades of writing in the margins of scorebooks has taught me that cricket's most reliable lie is the one told in the middle overs, when a team appears to be walking towards a total and nobody is asking what it is standing on.

Context: what actually happened at Newlands, and how many innings were played?

The series stood at 1-1. Cape Town's third Test was the decider, the first day was Tuesday. South Africa did not start badly. Aiden Markram (78) and Marques Ackerman (95 off 94) put on 140 for the first wicket. That was not just a partnership, it was a statement: the new ball spent, the pitch slow, the score 203/1.

Nothing in the live commentary at that stage was asking how many balls South Africa's numbers five, six and seven had faced in this series that actually moved. I don't know the answer — I don't have that data and I won't invent it. But what 203/1 conceals isn't a statistic, it's a structure: a strong opening stand postpones the audit of the rest of the batting order; it does not cancel it.

What the afternoon wrote down is this: 203/1 became 329 all out. That is nine wickets for 126 runs. The last four fell in six balls. Gerald Coetzee and Kagiso Rabada both returned golden ducks. Starc finished with 6/72, his best figures against South Africa, an improvement on Perth's 6/154.

And here come the graphics. The twentieth Test five-for. Add nine in ODIs and it's 29 in international cricket — level with Dale Steyn. Top of the active fast bowlers' list, ahead of Rabada (17) and Jasprit Bumrah (16). Sixth on the all-time list behind Richard Hadlee (41), Glenn McGrath (36), Waqar Younis (35), James Anderson (34) and Wasim Akram (31).

The table is tidy. That is its defect.

Core: two partnerships made 235; the other eight wickets shared 94

I watched the third session twice — once live, once with the scorecard open — to find the joins in the journey from 203/1 to 329. The answer is as clean as it is uncomfortable.

South Africa's 329 contains two large partnerships: 140 for the first wicket, and 95 for the seventh between Marco Jansen (52) and Ryan Rickelton (42). Between them, 235 runs. The other eight wickets produced 94 runs. While the stands were standing, South Africa were travelling at roughly 4.5 an over. When they broke, the wickets fell in clusters.

Starc's six wickets cost 72. South Africa's other four cost Australia 257. That single sentence reverses the whole five-for narrative. A bowler who genuinely breaks an innings leaves wickets scattered on both sides of the stand. Here there are two clean piles — a 140 and a 95 — with two hollow patches of 94 runs between them.

First hollow: David Bedingham made 50 and got out, Wiaan Mulder fell first ball for nought. Second hollow: the second new ball. Starc removed Rickelton for 42, bowled Jansen for 52, then had Coetzee and Rabada caught behind in consecutive golden ducks.

What has to be said about the second new ball is that it is not the bowler's reward, it is the captain's patience. Australia did not burn Starc in the first spell. They kept him in short bursts and put the ball in his hand once the seam was standing up again. Had he bowled twelve overs before lunch, that ball might not have found Coetzee's edge at all.

Thirty-two nights of cricket had already rewired how I watch football, but the real gift travelled the other way: I now read Test cricket through my powerplay–middle–death logic rather than football's data liturgy. And that logic says 203/1 was never a strong position. It was merely phase one, in which resources — the new ball, fresh quicks, bounce — were being spent with no interest coming back. In Test cricket the interest arrives in the shape of the second new ball. South Africa did not price it in for eighty overs.

Now the arithmetic that bothers me most.

The active fast bowlers' five-for table has Starc first on 20, from 210 innings. Rabada 17 from 136. Bumrah 16 from 99. Josh Hazlewood 15 from 151. Pat Cummins 14 from 141. Kemar Roach 11 from 156.

Divide that table by innings. How many innings each five-for actually took:

• Bumrah — one every 6.2 innings
• Rabada — one every 8.0
• Hazlewood — one every 10.1
• Cummins — one every 10.1
• Starc — one every 10.5
• Roach — one every 14.2

The table flips in six seconds: Starc isn't first on 20, he's fourth. What we are headlining as "top of the active fast bowlers" is a volume count, not an impact count. Bumrah has 16 five-fors in 99 innings. Run that rate across Starc's 210 innings and the arithmetic lands near 34 — Wasim Akram's neighbourhood. I am not claiming Bumrah is the better bowler. I am claiming that this list does not tell you who is most dangerous; it tells you who was handed the ball most often.

The second distortion sits in the all-format table, where Starc's 29 puts him level with Steyn at sixth. Nine of those 29 are ODI five-fors. Waqar's 35 contains 13 ODI hauls — 37 per cent of his total. Anderson's 34 contains two. The table is lining up men who were doing different jobs. Waqar's ODI five-fors came in the death overs of the 1990s, when five wickets in ten overs was a distinct craft. Anderson never played that role. Twenty-nine does not equal twenty-nine.

The empty IPL taught me to listen for what a crowd usually hides. In 2026, sixty matches without spectators showed why yorker usage climbed from 13.8 to 21.6 per cent — bowlers could finally hear the plan. I don't have that luxury at Newlands. There was a crowd, there was noise, and I don't have the stump-mic feed. So what I'm doing here is inference — but a more honest inference than the scorecard's, because it declares its limits.

The 2 a.m. thread outlived the match report, which is why I never stopped following it. At The Oval in 2026 India were bowled out for 158 and every desk filed the same story; the thread counted that the middle order had batted in eleven different configurations across 24 ODIs. At Newlands I see the same count running for the other team: South Africa lost eight wickets for 94 runs, and we are praising the bowler who took six of them.

Contrarian: here is where I could be wrong, and where it would be easy to prove it

The first thing that cuts against me: Starc created that collapse, he didn't just harvest it. The old-ball spell before the second new ball is the real story — Jansen and Rickelton were adding 95 and the run rate had to be throttled, or 400 would have made 6/72 meaningless. If Starc kept those twenty overs under 2.5 an over, the spell is genuinely great and my whole argument is just a well-turned line.

Second: the impact-per-ball model I use for auction autopsies does not transfer here. Before the 2026 mega auction I argued that ₹15.25 crore and ₹14 crore were buying availability, not impact — the ten most expensive buys had missed an average of 8.4 matches across the previous two seasons. In an auction the unit is clean: a match. Here the unit is a wicket, and a wicket is never solely the bowler's. It carries the batter's decision, the pitch's age, the captain's field, the weather. My translation breaks exactly there, and I would rather write that down than hide it.

Third: I watched this from Delhi in the morning, not from a dressing room. I have no pitch report, no hint from an Australian team meeting. What I did was read a scorecard and a replay three times.

So here is what would change my mind, stated in advance: if a pitch map shows Starc's line sitting on off stump far more in the second spell than the first, and if reverse swing was something he deliberately loaded once the ball aged, then this is a technical spell and not a batting failure. If that's proven, I'll withdraw the core of this piece. But not on the strength of the six-for — I want the line map, not the headline.

I treat every hot take as a question I failed to ask slowly. The question here is this: do we treat the five-for as a bowler's certificate because it is easy to count, or because it is right?

Takeaway: next time the number appears, look for the other count beside it

The series is 1-1 and the third Test decides it, so the next three days will be about whether Starc reaches 21. My prediction is different, and it is testable: this Test will be decided by whose number six and number seven can survive twenty overs in the second innings, not by who takes the next five-for. South Africa's first innings is already telling you why — they had two bridges, 140 and 95, and no foundation between them.

The other thing I'll be watching is Starc's first spell on the second morning, with the old ball, ten overs. If his economy sits under 2.5 without a wicket, that spell will matter more than this 6/72, even though nobody will ever count it. And if the Coetzee–Rabada golden ducks arrive again, I'll write exactly what I've written today.

The question isn't whether Starc is great. He is — twenty five-fors in 210 innings is not an accident. The question is whether we are measuring that greatness with the right number, or with the one that happened to be nearest to hand. South Africa gave away eight wickets for 94 at Newlands. Put 6/72 next to that count and you'll see where the fault lies — not in Starc's line, but in our table.