MONTMELO, Spain -- After 64 hours of track time spread over eight days at the Circuit de Catalunya, Ferrari and Mercedes were split by just 0.003 seconds on the final day of preseason testing. That is a remarkably small margin when you consider there has been a significant regulation change and a full winter of development since the two rivals last met on track, but look beyond the headline-grabbing times and the evidence suggests it is last year's runner up Ferrari that has moved ahead.
Speaking on Friday lunchtime before heading out to set his fastest time, reigning champion Lewis Hamilton estimated the gap to Ferrari was something closer to 0.5 seconds per lap. If true, that would be a disaster for Mercedes and represent the kind of performance deficit that could take months to recover -- if it could be recovered at all. But just like the 0.003 seconds figure on the timing screens, it was one of those testing numbers that creates instant headlines but struggles to stand up to deeper analysis.
Rather predictably, the true gap is somewhere in between -- but even the ballpark figures that follow in this article have weighty caveats attached. So far this year we have only seen the 2019 cars on one circuit, in relatively cold conditions and with incomplete information on important details such as fuel loads and engine settings. When you are dealing with a sport measured in fractions of a second, such factors make drawing conclusions ahead of the first race of the season difficult. Nevertheless, the past two weeks of testing did give us a rough guide of who's quickest, and at this stage of the year that's like gold dust for anyone wanting to know how the season will pan out.
The biggest problem with comparing single-lap performance in testing is that the times you see can be so easily skewed. What we want to compare is like-for-like, low-fuel, qualifying-style laps, but what we get are times set at different moments in the day, with different track conditions and no certainty over how much fuel was onboard. A car on the fastest compound tire, with its engine in its fastest mode can be up to 0.4 seconds shy of its potential just by carrying an extra 10 kilograms of fuel.
The practice of running with more fuel than necessary is colloquially known as "sandbagging", but it's not just about hiding performance. Heavier fuel loads also help expose nasty handling traits when the car is being pushed to the limit and that means a lap spent on the freshest, softest rubber can be slightly more valuable with a few extra kilos of fuel in the tank. There is the added advantage of keeping your rivals guessing, but ultimately testing is about understanding the car, not setting the fastest times.
"Testing you have to take with a pinch of salt," Hamilton said on Friday. "You could get to Australia and the gap you see could be bigger, you could get there and it could be less, you could get there and it could be equal. You really have no idea.
"There's no way when you're looking at GPS [data] to say what fuel load they [Ferrari] are on and what engine mode they're on. They're faster on the straights than us, for example. Now, is that because they have their flexi-wings that they've had in the past that drops drag? Or are they more turned up [on engine mode] than us, or are they just more efficient and just have less drag than us in general? Or are we heavier?
"So we have no idea. We really won't know until we get to the race."
It's also a truism of testing that the most impressive lap times are not always the fastest outright. Pirelli has five compounds of tire this year and the softest, known as the C5, yielded the fastest times over the past two weeks. But the C5 is a tire developed for use on street circuits and will not be on offer when F1 returns to the Circuit de Catalunya for the Spanish Grand Prix in May. The main problem is that by the end of a flying lap on the C5s, the rubber on the rear tires has a tendency to overheat and that can result in a loss of grip.
A look at the fastest sector times from Friday tells us that Hamilton was 0.2 seconds quicker than Vettel in the final sector, while the Ferrari was 0.2 seconds quicker than the Mercedes in sectors one and two. That anomalous final sector could be down to the characteristics of the two cars, but it is most likely linked to the time of day the laps were set. Hamilton benefitted from much cooler conditions in the final hour of testing that would have helped combat overheating, while Vettel had to cope with the heat of the midday sun.
So while Vettel and Hamilton both used the C5 to dip into the 1:16.2 seconds and set the two fastest times of the test, arguably their most representative flying laps came on the harder rubber that is less susceptible to overheating. Vettel clocked a 1:16.7 on the C3 compound just before lunch, which is two steps harder than the C5, and Hamilton set a 1:16.6 on the C4, which sits between the C3 and C5 in Pirelli's range.
Based on data amassed by Pirelli over the two weeks of testing, the average performance step between C5 and C4 was 0.6 seconds and the gap between C4 and C3 was also 0.6 seconds Therefore, if we "tire-correct" Vettel's lap time from C3 seconds to C5 seconds, it comes out somewhere in the region of a 1:15.4. Hamilton's C4 tires are only one step harder than a C5 and therefore he manages a 'tire-corrected' time of 1:16.0. Both lap times come out quicker than the driver's ultimate fastest laps on the C5 seconds, but once again the Ferrari holds an advantage.
However, it's worth adding the pinch of salt that Hamilton talked about at this stage. First off, Pirelli's estimated performance deltas between compounds are based on a spread of data across all teams, so if Mercedes and Ferrari sit at different ends of that data, they may or may not have been able to find the 0.6 seconds promised by the compound steps. Add to that the question of fuel loads and we are back to broad-brush figures, with the Mercedes seemingly slower than Ferrari but without a clear indication of the exact difference.
The only other point worth mentioning at this stage is how the cars looked on track. Again, Ferrari seemed to hold the advantage, with the SF90 planted and progressive compared to the Mercedes that still seemed to be lacking rear-end grip. Ferrari team principal Mattia Binotto went trackside while Vettel was setting his fastest laps, and on the whole he was impressed with the performance of Maranello's latest creation.
"It's always very interesting to go out and see the cars on track, because you can compare the different cars, the way they handle, brake, corner, acceleration," he said. "From what I could see, our car looked well balanced in all kinds of corners, slow, medium and high-speed ones. It looked stable under braking, allowing the drivers to attack the corners in a very aggressive way. I'd say that's what pleased me most from seeing our car out on track."
It wasn't the same story for Mercedes. More than once Hamilton had to correct for a sudden snap of oversteer in high-speed corners -- a trait that was also a concern for the team throughout the opening week of testing. A significant upgrade for the second week helped smother the car in more downforce, but the balance front to rear remained biased towards the nose. Although that is a clear negative for Mercedes' single-lap pace last week, if the team can find a way to make the car's handling more neutral by Melbourne, the extra confidence found by the drivers could be worth a couple of tenths in qualifying.
But as exciting and dramatic as the single-lap performance is, points are handed out for the race and not for qualifying. As a result, the majority of testing is focused on honing the car for a race distance and that means running heavier fuel. The first week is usually devoted to refining setup and is difficult to read, but by the second week teams attempt to simulate a full race distance.
In the battle between Mercedes and Ferrari, Hamilton was first to complete a race simulation on Wednesday afternoon of week two. However, the team left its imaginary lights-out relatively late, and the cool conditions in the late afternoon saw Hamilton struggle with front-left tire graining throughout. The problem proved disastrous for emulating a true race simulation and forced Hamilton onto a four-stop strategy over 66 laps.
Lessons learned, Mercedes put Valtteri Bottas on a race simulation much earlier in the afternoon on Thursday, which -- happily for those wishing to make comparisons -- coincided with Ferrari's first race simulation of the week with Charles Leclerc. This was as close as we got to a genuine like-for-like comparison throughout testing and it offered the most valuable data of all. The reason race simulations are so revealing is that they require a full (or nearly full) tank of fuel to complete, leaving a more faithful comparison between two cars and much less guesswork over fuel loads.
However, before we get stuck into the Ferrari/Mercedes comparison it is worth addressing Red Bull's absence from this analysis thus far. The RB15 appears to be in the running with the other two, but what happened during the race simulations on Thursday afternoon means useful lap time data is slim.
Like Bottas and Leclerc, Pierre Gasly was completing a race sim just after lunch, but unfortunately his ended in the wall. After making his first pit stop, the Frenchman got greedy on the entrance to Turn 9, dipped two wheels on the grass and spun off at 150 mph. The car was wrecked and, as a result, the race sim was a write off. For the record, however, Gasly had been lagging behind the pace of the Mercedes and Ferrari when the accident happened.
The consequences of the accident had a knock-on effect on Friday's running, when we had hoped to see more of the RB15's single lap pace. Verstappen suffered a gearbox failure after 29 laps, which he put down to the loss of two relatively fresh gearboxes over when Gasly was at the wheel. Verstappen's best effort by the end of the week was a 1:17.709 on C3 tyres, but clearly he felt more was possible.
"It's all related to what also happened yesterday [when Gasly crashed]," he said after completing just 29 laps on Friday. "It's just a shame, we know we're limited on parts. You're always limited with parts in testing and when you already have two which are destroyed, it won't be ideal."
Rewind to Thursday once more and the red flag interruption was a nuisance for Mercedes and Ferrari but didn't stop them completing their race sims. Both Leclerc and Bottas completed full, 66-lap race sims based on three stints on the C2 tyre. Removing anomalous laps where the cars dropped dramatically off the pace due to traffic or had to stop for Gasly's red flag, and the Ferrari was an average of 0.28 seconds per lap quicker than the Mercedes. That might not sound like much, but over a race distance it amounts to an 18-second advantage. In F1 terms that's pretty big and helps back up the suspicions from the one-lap pace that Ferrari holds an advantage of several tenths ahead of Melbourne.
Over the next week, both teams will work on refining setups and finalizing their race packages based on their testing data and that means the gap between the two will remain fluid. Perhaps Mercedes will make big gains by increasing rear downforce or perhaps the promise shown in Sector 3 will translate to a decent lap time through Albert Park's medium-speed corners. But barring any dramatic developments over the next two weeks, the smart money remains on the Ferrari, which has looked like the quickest car since day one.