Somewhat done...
Apparatus :- 1 Measuring jug, 1 small 'Acura' (don't ask !) ladies face cream jar - empty and cleansed with decent sealing screw top. 1 length of cotton, 2 stainless steel nuts, 1 Salter electronic scales. Approx 800 cc water.
Trap a few cm cotton in jar, lid on tightly, weight = 115 gms
Weigh jug and water = 1816 gms leave jug on scales.
Gently hang jar into water - result ! it only just sank, hold steady not touching insides of jug weight total 1926 gms.
Increase = 110 gms suggesting volume of jar is around 110 cc displacing 110 gms of water.
Jar shape more or less frustrum of cone having base diam. 3.1cm minor diam 2.7cm with a height of 4.3cm... Theoretical volume 113.8 cc
http://tinyurl.com/n2ymgg2
Error to volume actual measure (113.8 / 110) - 1 = 0.035 or + 3.5% - reasonable.
Just as a checker, placed two 30 gm s/s nuts in jar, closed lid tight dangled into water - steady state weight increase, as before 110 gms.
Additional weight of nuts had no effect (as expected) only volume displacement weight of water adding to overall weight.
Haven't tethered, but absolutely 100 % certain there can be no weight difference between t/tennis ball + flask + water with flask upright or inverted - cannot get something for nothing.
Here is the crux - how can, as depicted, one inversion weigh more than the other without adding or subtracting mass ?
When one realises they cannot - then one realises why the steel ball insertion causes its flask to descend or gain weight.
-0-