Begin with the native bimodal mixture, component means thirty five and sixty five. Initially each component has probability one half. Change the first component weight to zero point eight using its native weight slider. The second weight becomes zero point two, so probabilities still sum to one. The ideal mixture mean is the weighted sum of the component means. Initially seventeen point five plus thirty two point five gives fifty. After the action, twenty eight plus thirteen gives forty one. Each held bar uses sixteen pixels per mean unit, with contiguous component contributions. Increasing probability on the lower component pulls the ideal mean downward. This is the ideal Gaussian mixture expectation, not a promise about a finite random sample. Select the native Gaussian preset, and keep its centre at thirty five. Change standard deviation from six to twelve. In the source Gaussian formula, standard deviation appears both inside the exponential and in the normalization factor. Doubling spread makes the distribution wider. At the centre the exponential equals one, so its density is one divided by sigma times the square root of two pi. The centre density falls from approximately zero point zero six six four nine to zero point zero three three two five. These two held curves use the same absolute density scale and the same horizontal scale. The original tool adaptively rescales its chart height, so these teaching axes are explicitly different from native chart pixels. Probability remains the area under the ideal full curve, not its peak height. Return to the native bimodal mixture. Set both standard deviations to six and both weights to one half. Start with means thirty five and sixty five, then use the two native mean sliders to move the centres to forty five and fifty five. Each displayed curve is the exact weighted sum of the two source Gaussian densities, on common fixed axes. For this equal width, equal weight symmetric case, the midpoint changes curvature when centre separation equals twice sigma. Initial separation thirty exceeds twelve, giving two peaks. Final separation ten is below twelve, giving one central peak. That condition depends on the stated symmetry assumptions. The source labels modes using a separate heuristic threshold; we do not teach that threshold as a universal mathematical rule. These are ideal density comparisons, not physically measured data.