Are the visible leaders that crawl along the cloud base during spider lightning positive or negative or both? Mazur et al.,  suggested they were negative for a decaying thunderstorm in Florida. Optical observations using high-speed cameras suggest that most of those observed on the Great Plains may be positive polarity. However, negative leaders appear to propagate horizontally in-cloud (above cloud base) prior to the formation of the visible positive leaders that propagate below cloud base.
The video below shows four flashes from the backside of a linear MCS that passed over Rapid City, SD on 6/13/11. These four flashes exhibited the characteristics of spider lightning and correlated electric field data is being analyzed. I have annotated what I think is the negative and positive leader development during these flashes. Positive leaders that produce recoil leaders can be identified in standard-speed video as having detached (floating) leader segments and brightly forked tips. The features are actually integrated recoil leaders that occur during each of the 17 ms video images. A future post will discuss how positive leaders can be identified on standard-speed video and digital still images.
In 2008, I captured an impressive spider lightning flash that passed over Rapid City, SD on 6/25. A portion of this flash was recorded with a high-speed camera at 7,207 images per second (ips). Here is my analysis of the standard-speed video recording
As shown in the preceding video, a portion of the visible channels in the lower left side of the flash were recorded by a high-speed camera. Here is the video from that recording.
Below is a time-integrated (stacked) image from the high-speed video segment.
There are many recordings of spider lightning (also called anvil crawlers by storm chasers) on YouTube. Here are some of the better recordings. Note the similar pattern of preceding in-cloud brightness followed by apparent positive leader development.
I have personally reviewed the 1,000 ips high-speed video recorded by Dr. Mazur and although the quality suffered from compression, there does not appear to be any recoil leader activity or forked tips on the visible leaders that is characteristic of positive leaders. The pattern of branching by the leaders also resembles that of negative leaders. However, I cannot confidently say they are definitely negative leaders due to the quality of the recording and the fact that most of the recording had a large portion of the image saturated by the brightness of the flash. Dr. Marcelo Saba showed me a high-speed recording he captured, and there were clearly negative leaders visible just below cloud base. A positive leader developed after the negative leader passed and a +CG return stroke resulted. If spider flashes have extensive horizontal negative leader development that spatially precedes the visible positive leaders, then supporting electric field sensor data should indicate a negative field change (atmospheric electricity sign convention) due to the approach of the negative leaders. This should change to positive if positive leaders later approach and dominate the signal. However, in many of the standard-speed recordings, the positive leaders do not seem to travel as far as the preceding incloud activity.
Correlated observations using a LMA or interferometer and high-speed camera along with electric field sensors would likely show the relative location and timing of the negative leader development in-cloud (abundant spatially coherent LMA sources due to noisier propagation, visible in-cloud brightening and negative field change) and the positive leader development below cloud (less LMA sources with those produced by recoil leaders being spatially incoherent, visible recoil leaders and positive field change).
Furthermore, the horizontal leader development (both in-cloud and visible leaders) must be put in context of the entire flash they are associated with. Did they develop as part of an intracloud flash or ground flash? Where (and when relative to the spider) did any ground strokes associated with the flash occur? I have witnessed positive leaders propagating below cloud base that produce both +CG and -CG strokes. In the +CG case, a branch of the positive leader goes to ground. In the -CG case, the negative end of a recoil leader formed during positive leader development goes to ground. I will show and discuss examples of these to processes in a future post.
Mazur, V., X. Shao, and P. R. Krehbiel (1998), ‘‘Spider’’ lightning in intracloud and positive cloud-to-ground, J. Geophys. Res., 103(D16), 19,811 –19,822.