The ignition of natural combustible material by hot metal particles is an important fire ignition pathway by which wildland and wildland-urban-interface spot fires are started. There are numerous cases reported of wild fires started by clashing power-lines or from sparks generated by machines or engines. Similarly there are many cases reported of fires caused by grinding, welding and cutting sparks. Up to this point, research on hot particle spot fire ignition has largely focused on particle generation and transport. A small number of studies have examined what occurs after a hot particle contacts a natural fuel bed, but until recently the process remained poorly understood.This work describes an investigation of the effect of particle size, temperature and thermal properties on the ability of hot particles to cause flaming ignition of cellulosic fuel beds. Both experimental and theoretical approaches are used, with a focus on understanding the physics underlying the ignition process. For the experimental study, spheres of stainless steel, aluminum, brass and copper are heated in a tube furnace and dropped onto a powdered cellulose fuel bed; the occurrence of flaming ignition or lack thereof is visually observed and recorded. This procedure is repeated a large number of times for each metal type, varying particle diameter from 2 to 11 mm and particle temperature between 575 and 1100 degrees C. The results of these experiments are statistically analyzed to find approximate ignition boundaries and identify boundary trends with respect to the particle parameters of interest. Schlieren images recorded during the ignition experiments are also used to more accurately describe the ignition process. Based on these images, a simple theoretical model of hot particle spot fire ignition is developed and used to explore the experimental trends further. The model under-predicts the minimum ignition temperatures required for small spheres, but agrees qualitatively with the experimental data. Model simulations identify the important physics controlling ignition for different sized particles and clarify many of the experimental trends.The results show a hyperbolic relationship between particle size and temperature, with the larger particles requiring lower temperatures to ignite the cellulose than the smaller particles. For very small spheres, the temperature required for ignition is very sensitive to particle size, while for very large spheres, ignition temperature shows only a weak dependence on that variable. Flaming ignition of powdered cellulose by particles < 11 mm in size requires particle temperatures of at least 600C. Ignition has not been observed for 2 mm particles at temperatures up to 1100C, but the statistical analysis indicates that ignition by particles 2 mm and smaller may be possible at temperatures above 950C. No clear trend is observed with particle metal type, but copper particles require slightly higher ignition temperatures and seem more sensitive to experimental variation, likely due to their relatively high thermal conductivity. High-speed Schlieren images taken during the ignition experiments show that once particles land, they volatilize the powdered cellulose and the fuel vapor diffuses out into the surrounding air. Ignition occurs in the mixing layer between the vapor and the air, either during the initial expansion of the pyrolyzate away from the particle, or after a stable plume of volatiles has formed. Modeling results indicate that in the large-particle, high-conductivity limit, the particle's surface temperature remains close to its impact temperature over the timescales of ignition. As a result, particle thermal properties are unimportant and ignition occurs when heat generation in the mixing layer overcomes losses to the surrounding air. When the large-particle limit does not apply, the particle cools upon impact with the fuel bed. In addition to the losses to the surrounding air, the reaction zone experiences losses to the cooling particle and must generate a larger amount of heat for ignition to occur. Because cooling is so important, the initial bulk energy is more useful than impact temperature for predicting ignition by smaller particles. Along those lines, the additional heat of melting available to molten particles helps to resist particle cooling; as such, molten aluminum particles 3.5-7 mm in diameter can ignite at lower temperatures than solid particles of the same size with similar thermal properties. Decreasing volumetric heat capacity does increase minimum ignition temperature somewhat, but this effect is reduced for larger particles. Emissivity does not appear to have a significant effect on ignition propensity, suggesting that, over the timescales of ignition, radiation heat transfer is small relative to other modes of particle heat loss.