I'm not in agreement about 1.9 per, not enough data.

The manual says 2.38 gallons (U.S.) or 9 litres, and you get the fuel symbol and a chime. Now, that puts paid to the 1.9 per, unless they are suggesting you have almost half a gallon left at the zero bar point, and the last bar is 1.9 gallons...and Chipshot's data doesn't match that. We'll see as more data trickles in.

The manual says 2.38 gallons for what?

Let's do the math again, since I am certain about the 1.9 gallons per mark both as what it is supposed to be and what it is:

Assuming that both you and Chipshot fill the tank to the same height:

Reading Gallons to full Gallons in tank

F N/A unknown

7 3.15 unknown

6 N/A unknown

5 N/A unknown

H 8.75 unknown

3 N/A unknown

2 N/A unknown

1 14.75 unknown

E N/A unknown

(8.75-3.15)/3 = 1.87

(14.75-3.15)/6 = 1.94

(14.75-8.75)/3 = 2.00

Unless bars 6,5,3, and 2 are highly irregularly spaced, this is pretty close to 1.9 gallons per mark. The slope of the best line for the 3 data points is 1.94 gallons per mark. What we don't have to be able to fill in the gallons in tank column is the intercept because we don't know how many gallons really fit in the tank. However, if we assume 1.9 gallons per mark with an intercept of 0 gallons in the tank we get a nice alignment with all of the observations so far

Reading Gallons to full Gallons in tank Experimental

assuming 16.5 capacity Error (gallons)

F N/A 16.5 -

7 3.15 13.3 0.15

6 N/A 11.4 -

5 N/A 9.5 -

H 8.75 7.6 0.15

3 N/A 5.7 -

2 N/A 3.8 -

1 14.75 1.9 -0.15

E N/A 0 -

Gallons to full is the amount of fuel a properly setup automatic fill shutoff valve will put in the tank immediately after the fuel gauge has dropped to the noted mark. That is Chipshot put in 14.75 gallons when it dropped from 2 to 1 bars. Chipshot did not run it down to truly empty so we only have conjecture regarding how much fuel and how much usable fuel he has in his tank when he added 14.75 gallons.

Also, 1.9 gallons per mark isn't the number that fits the data, it is the number that fits Giulia's specification: 15.3/8 = 1.9125. It also happens to fit the data.

I expect this is a situation similar to AR stating that the 2.0T has 280HP. It actually does have about 280 WHP with presumably more at the flywheel of the engine. Likewise, I think that 15.3 gallons is the most AR says you can reliably use from the tank without significant risk of sucking air into the fuel pump rather than the actual amount of fuel that can fit in a properly filled tank or the amount of fuel that you can pump out of the tank while the car is perfectly level and stationary. However, the fuel gauge is telling you how much fuel is actually in the tank rather than how much you can reliably consume from the tank.