If you have a flashlight battery in your hand or sitting on the table, then usually the entire battery will be charged with respect to ground. In theory, the voltage on either battery terminal can be measured.
For example, one battery terminal may be at +715.5V with respect to earth, and the other terminal at +714.0V wrt earth (but still showing the expected 1.5V difference between the battery terminals.) Or, if it's sitting on a portable plastic table, and someone had earlier brushed their skin across that table, then the battery terminals may be at -18,000 and -18001.5 volts! If your hand then approaches the battery, those voltages may change drastically because of e-fields and capacitive-divider effects, yet the 1.5V between battery terminals remains constant.
In other words, the battery is actively maintaining the 1.5V between terminals. But also, the entire battery as a whole is acting as a capacitor-plate with a few picofarads, and the surface of nearby ground is supplying the second capacitor plate. Scuff your shoes on the carpet while you stand near the battery, and the battery voltage wrt earth will vary enormously. (However, standard DVM voltmeters cannot measure it. That 18KV which had been on the battery will vanish as soon as the grounded DVM touches one battery terminal.)
An ideal voltmeter can show these large voltages on the battery if one meter terminal is connected to earth. But a standard DVM voltmeter with 10meg input impedance will behave more like a short circuit, and instantly drain out the charge on the few-picofarads of battery surface-capacitance. Even though the battery terminals differ by 1.5V, the meter reads zero for either terminal. This occurs because the terminal-voltage is instantly altered by the 10meg-ohms of the meter, wrt ground. The meter discharges each battery terminal that it touches. (The other "un-touched" terminal then changes too, maintaining the constant 1.5V between terminals.)
How "instantly" does the meter alter the battery terminal's voltage?
Suppose the entire battery acts as a 2pF capacitor plate, with ground being the other plate. If we try to measure the voltage on a single battery terminal, our DVM connects a 10M resistor between that terminal and ground. The measured voltage will fall to zero, with a time constant of:
Tc = RC
Tc = 10^7 ohms * 2pF
Tc = 20 microseconds
The moral of the story: in circuits, voltage is measured between two points on the circuit, not one. If you try to measure the voltage at a single point wrt ground, you'll end up with crazy enormous values which are strongly influenced by nearby static-charged objects. (But also you'll need a specialized voltmeter to measure these values at all. 10M ohms is far too small. Instead use an "electrometer" voltmeter with 10^20 ohms input impedance.)